ETSI DVB Measurement Guide

For full format paper (including table, picture, graphic etc) Please open at  ETSI DVB Measurement Guide



ETSI TR 101 290 V1.2.1 (2001-05)
Technical Report
Digital Video Broadcasting (DVB);
Measurement guidelines for DVB systems
European Broadcasting Union Union Européenne de Radio-Télévision
EBU·UER
ETSI
2 ETSI TR 101 290 V1.2.1 (2001-05)
Reference
RTR/JTC-DVB-77
Keywords
Broadcasting, digital, video, DVB, TV
ETSI
650 Route des Lucioles
F-06921 Sophia Antipolis Cedex - FRANCE
Tel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16
Siret N° 348 623 562 00017 - NAF 742 C
Association à but non lucratif enregistrée à la
Sous-Préfecture de Grasse (06) N° 7803/88
Important notice
Individual copies of the present document can be downloaded from:
http://www.etsi.org
The present document may be made available in more than one electronic version or in print. In any case of existing or
perceived difference in contents between such versions, the reference version is the Portable Document Format (PDF).
In case of dispute, the reference shall be the printing on ETSI printers of the PDF version kept on a specific network drive
within ETSI Secretariat.
Users of the present document should be aware that the document may be subject to revision or change of status.
Information on the current status of this and other ETSI documents is available at http://www.etsi.org/tb/status/
If you find errors in the present document, send your comment to:
editor@etsi.fr
Copyright Notification
No part may be reproduced except as authorized by written permission.
The copyright and the foregoing restriction extend to reproduction in all media.
© European Telecommunications Standards Institute 2001.
© European Broadcasting Union 2001.
All rights reserved.
ETSI
3 ETSI TR 101 290 V1.2.1 (2001-05)
Contents
Intellectual Property Rights ........................................................................................................................10
Foreword...................................................................................................................................................10
1 Scope...............................................................................................................................................11
2 References .......................................................................................................................................11
3 Definitions and abbreviations............................................................................................................12
3.1 Definitions ............................................................................................................................................... 12
3.2 Abbreviations........................................................................................................................................... 12
4 General............................................................................................................................................14
5 Measurement and analysis of the MPEG-2 Transport Stream............................................................16
5.1 General .................................................................................................................................................... 16
5.2 List of parameters recommended for evaluation ......................................................................................... 16
5.2.1 First priority: necessary for de-codability (basic monitoring)................................................................. 17
5.2.2 Second priority: recommended for continuous or periodic monitoring................................................... 19
5.2.3 Third priority: application dependant monitoring.................................................................................. 20
5.3 Measurement of MPEG-2 Transport Streams in networks .......................................................................... 25
5.3.1 Introduction ........................................................................................................................................ 25
5.3.2 System clock and PCR measurements .................................................................................................. 25
5.3.2.1 Reference model for system clock and PCR measurements.............................................................. 25
5.3.2.2 Measurement descriptions............................................................................................................... 27
5.3.2.3 Program Clock Reference - Frequency Offset PCR_FO................................................................... 28
5.3.2.4 Program Clock Reference – Drift Rate PCR_DR............................................................................. 28
5.3.2.5 Program Clock Reference - Overall Jitter PCR_OJ .......................................................................... 29
5.3.2.6 Program Clock Reference – Accuracy PCR_AC.............................................................................. 29
5.3.3 Bitrate measurement ............................................................................................................................ 29
5.3.3.1 Bitrate measurement algorithm ....................................................................................................... 30
5.3.3.2 Preferred values for Bitrate Measurement........................................................................................ 31
5.3.3.3 Nomenclature................................................................................................................................ 31
5.3.4 Consistency of information check......................................................................................................... 32
5.3.4.1 Transport_Stream_ID check............................................................................................................ 32
5.3.5 TS parameters in transmission systems with reduced SI data................................................................. 32
5.4 Measurement of availability at MPEG-2 Transport Stream level................................................................. 33
5.5 Evaluation of service performance by combination of TS related parameters .............................................. 34
5.5.1 Service_Availability_Error and Service_ Availability _Error_Ratio ...................................................... 35
5.5.2 Service_Degradation_Error and Service_Degradation_Error_Ratio....................................................... 35
5.5.3 Service_Impairments_Error and Service_Impairments_Error_Ratio...................................................... 36
5.6 Parameters for CI related applications........................................................................................................ 36
5.6.1 Latency............................................................................................................................................... 37
5.6.2 CI_module_delay_variation ................................................................................................................. 38
5.6.3 Input_output_TS comparison ............................................................................................................... 38
5.6.4 CI_module_throughput ........................................................................................................................ 38
5.6.5 Valid TS on CI.................................................................................................................................... 38
6 Common parameters for satellite and cable transmission media.........................................................39
6.1 System availability ................................................................................................................................... 39
6.2 Link availability ....................................................................................................................................... 39
6.3 BER before RS decoder............................................................................................................................ 40
6.3.1 Out of service ..................................................................................................................................... 40
6.3.2 In service ............................................................................................................................................ 40
6.4 Error events logging ................................................................................................................................. 40
6.5 Transmitter symbol clock jitter and accuracy ............................................................................................. 41
6.6 RF/IF signal power ................................................................................................................................... 41
6.7 Noise power ............................................................................................................................................. 41
6.8 Bit error count after RS............................................................................................................................. 42
6.9 IQ signal analysis ..................................................................................................................................... 42
ETSI
4 ETSI TR 101 290 V1.2.1 (2001-05)
6.9.1 Introduction ........................................................................................................................................ 42
6.9.2 Modulation Error Ratio (MER) ............................................................................................................ 43
6.9.3 System Target Error (STE)................................................................................................................... 44
6.9.4 Carrier suppression .............................................................................................................................. 45
6.9.5 Amplitude Imbalance (AI).................................................................................................................... 45
6.9.6 Quadrature Error (QE) ......................................................................................................................... 45
6.9.7 Residual Target Error (RTE) ................................................................................................................ 46
6.9.8 Coherent interferer ............................................................................................................................... 46
6.9.9 Phase Jitter (PJ)................................................................................................................................... 47
6.9.10 Signal-to-Noise Ratio (SNR) ................................................................................................................ 48
6.10 Interference.............................................................................................................................................. 48
7 Cable specific measurements ............................................................................................................49
7.1 Noise margin............................................................................................................................................ 49
7.2 Estimated noise margin............................................................................................................................. 49
7.3 Signal quality margin test .......................................................................................................................... 49
7.4 Equivalent Noise Degradation (END) ........................................................................................................ 50
7.5 BER vs. Eb/N0.......................................................................................................................................... 52
7.6 Phase noise of RF carrier ........................................................................................................................... 52
7.7 Amplitude, phase and impulse response of the channel............................................................................... 52
7.8 Out of band emissions ............................................................................................................................... 53
8 Satellite specific measurements.........................................................................................................53
8.1 BER before Viterbi decoding..................................................................................................................... 53
8.2 Receive BER vs. Eb/No ............................................................................................................................. 53
8.3 IF spectrum.............................................................................................................................................. 54
9 Measurements specific for a terrestrial (DVB-T) system ...................................................................54
9.1 RF frequency measurements..................................................................................................................... 56
9.1.1 RF frequency accuracy (Precision) ....................................................................................................... 56
9.1.2 RF channel width (Sampling Frequency Accuracy)............................................................................... 57
9.1.3 Symbol Length measurement at RF (Guard Interval verification) .......................................................... 58
9.2 Selectivity................................................................................................................................................ 58
9.3 AFC capture range ................................................................................................................................... 58
9.4 Phase noise of Local Oscillators (LO) ........................................................................................................ 58
9.5 RF/IF signal power ................................................................................................................................... 59
9.6 Noise power ............................................................................................................................................. 60
9.7 RF and IF spectrum.................................................................................................................................. 60
9.8 Receiver sensitivity/dynamic range for a Gaussian channel ........................................................................ 60
9.9 Equivalent Noise Degradation (END) ........................................................................................................ 60
9.9.1 Equivalent Noise Floor (ENF).............................................................................................................. 61
9.10 Linearity characterization (shoulder attenuation) ........................................................................................ 62
9.11 Power efficiency....................................................................................................................................... 62
9.12 Coherent interferer ................................................................................................................................... 62
9.13 BER vs. C/N ratio by variation of transmitter power .................................................................................. 62
9.14 BER vs. C/N ratio by variation of Gaussian noise power ............................................................................ 63
9.15 BER before Viterbi (inner) decoder ........................................................................................................... 63
9.16 BER before RS (outer) decoder.................................................................................................................. 64
9.16.1 Out of Service..................................................................................................................................... 64
9.16.2 In Service ........................................................................................................................................... 64
9.17 BER after RS (outer) decoder (Bit error count) .......................................................................................... 64
9.18 IQ signal analysis ..................................................................................................................................... 65
9.18.1 Introduction ........................................................................................................................................ 65
9.18.2 Modulation Error Ratio (MER) ............................................................................................................ 66
9.18.3 System Target Error (STE)................................................................................................................... 66
9.18.4 Carrier Suppression (CS)...................................................................................................................... 67
9.18.5 Amplitude Imbalance (AI).................................................................................................................... 68
9.18.6 Quadrature Error (QE) ......................................................................................................................... 69
9.18.7 Phase Jitter (PJ)................................................................................................................................... 69
9.19 Overall signal delay.................................................................................................................................. 71
9.20 SFN synchronization ................................................................................................................................ 72
9.20.1 MIP_timing_error ............................................................................................................................... 72
ETSI
5 ETSI TR 101 290 V1.2.1 (2001-05)
9.20.2 MIP_structure_error............................................................................................................................ 73
9.20.3 MIP_presence_error............................................................................................................................ 73
9.20.4 MIP_pointer_error ............................................................................................................................... 74
9.20.5 MIP_periodicity_error.......................................................................................................................... 74
9.20.6 MIP_ts_rate_error ............................................................................................................................... 75
9.21 System Error Performance........................................................................................................................ 76
10 Recommendations for the measurement of delays in DVB systems ...................................................76
10.1 Introduction.............................................................................................................................................. 76
10.2 Technical description of the measurements ................................................................................................ 77
10.2.1 Definition of input signal...................................................................................................................... 77
10.2.2 Overall delay and end-to-end encoder delay.......................................................................................... 78
10.2.2.1 Measurement of overall delay ......................................................................................................... 78
10.2.2.2 Measurement of end to end encoder delay....................................................................................... 79
10.2.2.3 Total decoder delay measurement. .................................................................................................. 79
10.2.2.4 Measurement of Relative Audio/Video delay - Lip Sync ................................................................. 80
Annex A (informative): General measurement methods .................................................................81
A.1 Introduction.....................................................................................................................................81
A.2 Null packet definition .......................................................................................................................81
A.3 Description of the procedure for "Estimated Noise Margin" by applying statistical analysis on the
constellation data ..............................................................................................................................82
A.4 Set-up for RF phase noise measurements using a spectrum analyser ..................................................83
A.5 Amplitude, phase and impulse response of the channel......................................................................84
A.6 Out of band emissions......................................................................................................................85
Annex B (informative): Examples for test set-ups for satellite and cable systems..........................86
B.1 System availability ...........................................................................................................................86
B.2 Link availability...............................................................................................................................86
B.3 BER before RS................................................................................................................................86
B.3.1 Out of service measurement...................................................................................................................... 87
B.3.2 In service measurement ............................................................................................................................. 87
B.4 Event error logging..........................................................................................................................87
B.5 Transmitter symbol clock jitter and accuracy ....................................................................................88
B.6 RF/IF signal power ...........................................................................................................................88
B.7 Noise power ....................................................................................................................................88
B.7.1 Out of service measurement...................................................................................................................... 88
B.7.2 In service measurement ............................................................................................................................. 89
B.8 BER after RS...................................................................................................................................89
B.9 I/Q signal analysis ............................................................................................................................89
B.10 Service data rate measurement ..........................................................................................................89
B.11 Noise margin ...................................................................................................................................89
B.11.1 Recommended equipment......................................................................................................................... 90
B.11.2 Remarks and precautions.......................................................................................................................... 90
B.11.3 Measurement procedure............................................................................................................................ 91
ETSI
6 ETSI TR 101 290 V1.2.1 (2001-05)
B.12 Equivalent Noise Degradation (END) ...............................................................................................91
B.13 BER vs. Eb/N0 .................................................................................................................................92
B.14 Equalizer specification.....................................................................................................................92
B.15 BER before Viterbi decoding ............................................................................................................93
B.16 Receive BER vs. Eb/N0...................................................................................................................93
B.17 IF spectrum......................................................................................................................................94
Annex C (informative): Measurement parameter definition...........................................................95
C.1 Definition of Vector Error Measures .................................................................................................95
C.2 Comparison between MER and EVM ...............................................................................................95
C.3 Conclusions regarding MER and EVM.............................................................................................96
Annex D (informative): Exact values of BER vs. Eb/N0 for DVB-C systems...................................97
Annex E (informative): Examples for the terrestrial system test set-ups........................................98
E.1 RF frequency accuracy .............................................................................................................................. 98
E.1.1 Frequency measurements in DVB-T .......................................................................................................... 98
E.1.2 Measurement in other cases ......................................................................................................................100
E.1.3 Calculation of the external pilots frequency when they do not have continual phase...................................101
E.1.4 Measuring the symbol length and verifying the Guard Interval ..................................................................105
E.1.5 Measuring the occupied bandwidth, and calculation of the frequency spacing and sampling frequency ......108
E.2 Selectivity......................................................................................................................................108
E.3 AFC capture range..........................................................................................................................108
E.4 Phase noise of Local Oscillators (LO) .......................................................................................................109
E.4.1 Practical information on phase noise measurements ..................................................................................109
E.5 RF/IF signal power ........................................................................................................................110
E.5.1 Procedure 1 (power metre)........................................................................................................................110
E.5.2 Procedure 2 (spectrum analyser) ...............................................................................................................111
E.6 Noise power ..................................................................................................................................111
E.6.1 Procedure 1.............................................................................................................................................111
E.6.2 Procedure 2.............................................................................................................................................111
E.6.3 Procedure 3.............................................................................................................................................111
E.6.4 Measurement of noise with a spectrum analyser ........................................................................................112
E.7 RF and IF spectrum ........................................................................................................................112
E.8 Receiver sensitivity/dynamic range for a Gaussian channel .............................................................113
E.9 Equivalent Noise Degradation (END) .............................................................................................113
E.9.1 Description of the measurement method for END.....................................................................................113
E.9.2 Conversion method between ENF and END..............................................................................................114
E.10 Linearity characterization (shoulder attenuation) .............................................................................115
E.10.1 Equipment...............................................................................................................................................115
E.10.2 Remarks and precautions.........................................................................................................................115
E.10.3 Measurement procedure (example for UHF channel 47)............................................................................116
ETSI
7 ETSI TR 101 290 V1.2.1 (2001-05)
E.11 Power efficiency............................................................................................................................117
E.12 Coherent interferer.........................................................................................................................117
E.13 BER vs. C/N by variation of transmitter power ...............................................................................117
E.14 BER vs. C/N by variation of Gaussian noise power .........................................................................118
E.15 BER before Viterbi (inner) decoder.................................................................................................118
E.16 Overall signal delay ........................................................................................................................118
Annex F (informative): Specification of test signals of DVB-T modulator ...................................121
F.1 Introduction...................................................................................................................................121
F.2 Input signal....................................................................................................................................121
F.3 Test modes ....................................................................................................................................122
F.4 Test points .....................................................................................................................................122
F.5 File format for interchange of simulated data ..................................................................................122
F.5.1 Test point number....................................................................................................................................123
F.5.2 Length of data buffer ...............................................................................................................................123
F.5.3 Bit ordering after inner interleaver ............................................................................................................123
F.5.4 Carrier allocation.....................................................................................................................................123
F.5.5 Scaling....................................................................................................................................................124
F.5.6 Constellation ...........................................................................................................................................124
F.5.7 Hierarchy................................................................................................................................................124
F.5.8 Code rate LP and HP...............................................................................................................................124
F.5.9 Guard interval .........................................................................................................................................125
F.5.10 Transmission mode..................................................................................................................................125
F.5.11 Data format .............................................................................................................................................125
F.5.12 Example..................................................................................................................................................125
ETSI
8 ETSI TR 101 290 V1.2.1 (2001-05)
Annex G (informative): Theoretical background information on measurement techniques........127
G.1 Overview.......................................................................................................................................127
G.2 RF/IF power ("carrier")...................................................................................................................127
G.3 Noise level.....................................................................................................................................128
G.4 Energy-per-bit (Eb) .........................................................................................................................129
G.5 C/N ratio and Eb/No ratio ................................................................................................................129
G.6 Practical application of the measurements ....................................................................................... 129
G.7 Example ........................................................................................................................................130
G.8 Signal-to-Noise Ratio (SNR) and Modulation Error Ratio (MER) ...................................................132
G.9 BER vs. C/N..................................................................................................................................132
G.10 Error probability of Quadrature Amplitude Modulation (QAM) ......................................................133
G.11 Error probability of QPSK ..............................................................................................................134
G.12 Error probability after Viterbi decoding ..........................................................................................135
G.13 Error probability after RS decoding.................................................................................................135
G.14 BEP vs. C/N for DVB cable transmission........................................................................................136
G.15 BER vs. C/N for DVB satellite transmission ...................................................................................137
G.16 Adding noise to a noisy signal.........................................................................................................138
Annex H: Void ...............................................................................................................................141
Annex I (informative): PCR related measurements .....................................................................142
I.1 Introduction...................................................................................................................................142
I.2 Limits ............................................................................................................................................142
I.3 Equations.......................................................................................................................................143
I.4 Mask .............................................................................................................................................144
I.5 Break frequencies ...........................................................................................................................145
I.6 Further implicit limitations..............................................................................................................146
I.7 Measurement procedures ................................................................................................................147
I.7.1 PCR_Accuracy (PCR_AC) .......................................................................................................................148
I.7.2 PCR_drift_rate (PCR_DR) .......................................................................................................................149
I.7.3 PCR_frequency_offset (PCR_FO) ............................................................................................................150
I.7.4 PCR_overall_jitter Measurement ..............................................................................................................150
I.8 Considerations on performing PCR measurements ..........................................................................151
I.9 Choice of filters in PCR measurement.............................................................................................152
I.9.1 Why is there a choice ?.............................................................................................................................152
I.9.2 Higher demarcation frequencies................................................................................................................153
I.9.3 Lower demarcation frequencies ................................................................................................................154
I.10 Excitation model for PCR measurement devices .............................................................................154
I.10.1 Introduction.............................................................................................................................................154
I.10.2 Constraints on the definition of a stream...................................................................................................155
I.10.3 The Algorithm.........................................................................................................................................158
I.10.3.1 Parameterization ................................................................................................................................159
I.10.3.2 Scheduling.........................................................................................................................................159
I.10.3.3 Synthesis ...........................................................................................................................................159
I.10.4 The Pseudo-C code..................................................................................................................................159
ETSI
9 ETSI TR 101 290 V1.2.1 (2001-05)
I.10.5 Parameter definitions and example values.................................................................................................162
Annex J (informative): Bitrate related measurements..................................................................164
J.1 Introduction...................................................................................................................................164
J.1.1 Purpose of bitrate measurement ................................................................................................................164
J.1.2 User Rate versus Multiplex Rate...............................................................................................................164
J.1.3 User rate applications ...............................................................................................................................166
J.2 Principles of Bit rate measurement ..................................................................................................166
J.2.1 Gate or Window function .........................................................................................................................166
J.2.2 "Continuous window"..............................................................................................................................167
J.2.3 Time Gate values:....................................................................................................................................167
J.2.4 Rate measurements in a transport stream...................................................................................................167
J.3 Use of the MG profiles ...................................................................................................................168
J.3.1 MGB1 Profile - the backwards compatible profile.....................................................................................168
J.3.2 MGB2 Profile - the Basic bitrate profile....................................................................................................168
J.3.3 MGB3 Profile - the precise Peak bitrate profile .........................................................................................168
J.3.4 MGB4 Profile - the precise profile ............................................................................................................168
J.3.5 MGB5 Profile - the user profile ................................................................................................................168
J.4 Error values in the measurements....................................................................................................169
J.4.1 Very Precise measurements ......................................................................................................................170
Annex K (informative): DVB-T channel characteristics................................................................171
K.1 Theoretical channel profiles for simulations without Doppler shift ..................................................171
K.2 Profiles for realtime simulations without Doppler shift....................................................................172
K.3 Profiles for realtime simulation with Doppler shift (mobile channel simulation) ..............................173
Annex L (informative): Bibliography............................................................................................174
History ....................................................................................................................................................175
ETSI
10 ETSI TR 101 290 V1.2.1 (2001-05)
Intellectual Property Rights
IPRs essential or potentially essential to the present document may have been declared to ETSI. The information
pertaining to these essential IPRs, if any, is publicly available for ETSI members and non-members, and can be found
in ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to ETSI in
respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the ETSI Web
server (http://www.etsi.org/ipr).
Pursuant to the ETSI IPR Policy, no investigation, including IPR searches, has been carried out by ETSI. No guarantee
can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on the ETSI Web
server) which are, or may be, or may become, essential to the present document.
Foreword
This Technical Report (TR) has been produced by Joint Technical Committee (JTC) Broadcast of the European
Broadcasting Union (EBU), Comité Européen de Normalisation ELECtrotechnique (CENELEC) and the European
Telecommunications Standards Institute (ETSI).
NOTE: The EBU/ETSI JTC Broadcast was established in 1990 to co-ordinate the drafting of standards in the
specific field of broadcasting and related fields. Since 1995 the JTC Broadcast became a tripartite body
by including in the Memorandum of Understanding also CENELEC, which is responsible for the
standardization of radio and television receivers. The EBU is a professional association of broadcasting
organizations whose work includes the co-ordination of its members' activities in the technical, legal,
programme-making and programme-exchange domains. The EBU has active members in about 60
countries in the European broadcasting area; its headquarters is in Geneva.
European Broadcasting Union
CH-1218 GRAND SACONNEX (Geneva)
Switzerland
Tel: +41 22 717 21 11
Fax: +41 22 717 24 81
Founded in September 1993, the DVB Project is a market-led consortium of public and private sector organizations in
the television industry. Its aim is to establish the framework for the introduction of MPEG-2 based digital television
services. Now comprising over 200 organizations from more than 25 countries around the world, DVB fosters
market-led systems, which meet the real needs, and economic circumstances, of the consumer electronics and the
broadcast industry.
ETSI
11 ETSI TR 101 290 V1.2.1 (2001-05)
1 Scope
The present document provides guidelines for measurement in Digital Video Broadcasting (DVB) satellite, cable and
terrestrial and related digital television systems. The present document defines a number of measurement techniques,
such that the results obtained are comparable when the measurement is carried out in compliance with the appropriate
definition.
The present document uses terminology used in EN 300 421 [5], EN 300 429 [6], EN 300 468 [7] and EN 300 744 [9]
and it should be read in conjunctions with them.
2 References
For the purposes of this Technical Report (TR), the following references apply:
[1] ISO/IEC 13818-1 (ITU-T Recommendation H.222.0): "Information technology - Generic coding
of moving pictures and associated audio information: Systems".
[2] ISO/IEC 13818-4: "Information technology - Generic coding of moving pictures and associated
audio information - Part 4: Conformance testing ".
[3] ISO/IEC 13818-9: "Information technology - Generic coding of moving pictures and associated
audio information - Part 9: Extension for real time interface for systems decoders".
[4] ETSI TR 101 154: "Digital Video Broadcasting (DVB); Implementation guidelines for the use of
MPEG-2 Systems, Video and Audio in satellite, cable and terrestrial broadcasting applications".
[5] ETSI EN 300 421: "Digital Video Broadcasting (DVB); Framing structure, channel coding and
modulation for 11/12 GHz satellite services".
[6] ETSI EN 300 429: "Digital Video Broadcasting (DVB); Framing structure, channel coding and
modulation for cable systems".
[7] ETSI EN 300 468: "Digital Video Broadcasting (DVB); Specification for Service Information (SI)
in DVB systems".
[8] ETSI TR 101 211: "Digital Video Broadcasting (DVB); Guidelines on implementation and usage
of Service Information (SI)".
[9] ETSI EN 300 744: "Digital Video Broadcasting (DVB); Framing structure, channel coding and
modulation for digital terrestrial television".
[10] EN 50083-9: "Cable networks for television signals, sound signals and interactive services -
Part 9: Interfaces for CATV/SMATV headends and similar professional equipment for
DVB/MPEG-2 transport streams".
[11] ITU-T Recommendation G.826: "Error performance parameters and objectives for international,
constant bit rate digital paths at or above the primary rate".
[12] ITU-T Recommendation O.151: "Error performance measuring equipment operating at the
primary rate and above".
[13] ETSI EN 300 473: "Digital Video Broadcasting (DVB); Satellite Master Antenna Television
(SMATV) distribution systems".
[14] ETSI TS 101 191: "Digital Video Broadcasting (DVB); DVB mega-frame for Single Frequency
Network (SFN) synchronization".
[15] ETSI EN 300 748: "Digital Video Broadcasting (DVB); Multipoint Video Distribution Systems
(MVDS) at 10 GHz and above".
[16] ETSI EN 300 749: "Digital Video Broadcasting (DVB); Microwave Multipoint Distribution
Systems (MMDS) below 10 GHz".
ETSI
12 ETSI TR 101 290 V1.2.1 (2001-05)
[17] ISO 639: "Code for the representation of names of languages ".
[18] ETSI EN 301 210: "Digital Video Broadcasting (DVB); Framing structure, channel coding and
modulation for Digital Satellite News Gathering (DSNG) and other contribution applications by
satellite".
[19] ETSI ETS 300 813: "Digital Video Broadcasting (DVB); DVB interfaces to Plesiochronous
Digital Hierarchy (PDH) networks".
[20] ETSI ETS 300 814: "Digital Video Broadcasting (DVB); DVB interfaces to Synchronous Digital
Hierarchy (SDH) networks".
[21] ETSI ETR 290: "Digital Video Broadcasting (DVB); Measurement guidelines for DVB systems".
[22] ISO/IEC 13818 series: "Information Technology - Generic coding of moving pictures and
associated audio information".
[23] EN 50221: "Common interface specification for conditional access and other digital video
broadcasting decoder applications".
3 Definitions and abbreviations
3.1 Definitions
For the purposes of the present document, the following terms and definitions apply:
MPEG-2: Refers to the ISO/IEC 13818 [22] series. Systems coding is defined in part 1. Video coding is defined in part
2. Audio coding is defined in part 3.
multiplex: stream of all the digital data carrying one or more services within a single physical channel
Service Information (SI): digital data describing the delivery system, content and scheduling/timing of broadcast data
streams, etc.
It includes MPEG-2 Program Specific Information (PSI) together with independently defined extensions.
Transport Stream (TS): Data structure defined in ISO/IEC 13818-1 [1]. It is the basis of the Digital Video
Broadcasting (DVB) related standards.
3.2 Abbreviations
For the purposes of the present document, the following abbreviations apply:
AFC Automatic Frequency Control
AI Amplitude Imbalance
ASCII American Standard Code for Information Interchange
ATM Asynchronous Transfer Mode
AWGN Additive White Gaussian Noise
BAT Bouquet Association Table
BEP Bit Error Probability
BER Bit Error Rate
bslbf bit string, left bit first
BW BandWidth
C/N ratio of RF or IF signal power to noise power
CA Conditional Access
CATV Community Antenna TeleVision
CPE Common Phase Error
CRC Cyclic Redundancy Check
CS Carrier Suppression
CSO Composite Second Order
CTB Composite Triple Beat
ETSI
13 ETSI TR 101 290 V1.2.1 (2001-05)
CW Continuous Wave
DC Direct Current
DVB Digital Video Broadcasting
DVB-C Digital Video Broadcasting baseline system for digital cable television (EN 300 429 [6])
DVB-CS Digital Video Broadcasting baseline system for SMATV distribution systems (EN 300 473 [13])
DVB-MC Digital Video Broadcasting baseline system for Multi-point Video Distribution Systems below 10
GHz (EN 300 749 [16])
DVB-MS Digital Video Broadcasting baseline system for Multi-point Video Distribution Systems at 10 GHz
and above (EN 300 748 [15])
DVB-S Digital Video Broadcasting baseline system for digital satellite television (EN 300 421 [5])
DVB-T Digital Video Broadcasting baseline system for digital terrestrial television (EN 300 744 [9])
EB Errored Block
EIT Event Information Table
EMM Entitlement Management Message
ENB Equivalent Noise Bandwidth
END Equivalent Noise Degradation
ES Errored Second
ETR ETSI Technical Report
ETS European Telecommunication Standard
EVM Error Vector Magnitude
FEC Forward Error Correction
FFT Fast Fourier Transform
HEX Hexadecimal
HPF High Pass Filter
ICI Inter-Carrier Interference
IEC International Electrotechnical Commission
IF Intermediate Frequency
IFFT Inverse FFT (Fast Fourier Transform)
IQ In-phase/Quadrature components
IRD Integrated Receiver Decoder
ISO International Organization for Standardization
ITU International Telecommunication Union
LAT Link Available Time
LO Local Oscillator
LPF Low Pass Filter
MER Modulation Error Ratio
MIP Mega-frame Initialization Packet
MMDS Microwave Multi-point Distribution Systems (or Multi-channel Multi-point Distribution Systems)
MPEG Moving Picture Experts Group
MVDS Multi-point Video Distribution Systems
NIT Network Information Table
OFDM Orthogonal Frequency Division Multiplex
PAT Program Association Table
PCR Program Clock Reference
PE Phase Error
PID Packet Identifier
PJ Phase Jitter
PLL Phase Locked Loop
PMT Program Map Table
PRBS Pseudo Randon Binary Sequence
printf symbol in the C programming language
PSI MPEG-2 Program Specific Information (as defined in ISO/IEC 13818-1 [1])
PTS Presentation Time Stamps
QAM Quadrature Amplitude Modulation
QE Quadrature Error
QEF Quasi Error Free
QEV Quadrature Error Vector
QPSK Quaternary Phase Shift Keying
RF Radio Frequency
RMS Root Mean Square
RS Reed-Solomon
RST Running Status Table (see EN 300 468 [7])
ETSI
14 ETSI TR 101 290 V1.2.1 (2001-05)
RTE Residual Target Error
SDP Severely Disturbed Period
SDT Service Description Table
SEP Symbol Error Probability
SER Symbol Error Rate
SES Seriously Errored Second
SFN Single Frequency Network
SI Service Information
SMATV Satellite Master Antenna TeleVision
SNR Signal-to-Noise Ratio
STD System Target Decoder
STE System Target Error
STED STE Deviation
STEM STE Mean
TDT Time and Date Table
TEV Target Error Vector
TOT Time Offset Table
TPS Transmission Parameter Signalling
TS Transport Stream
TV TeleVision
UI Unit Interval
uimsbf unsigned integer, most significant bit first
UTC Universal Time Co-ordinated
4 General
The Digital Video Broadcasting (DVB) set of digital TV standards specify baseline systems for various transmission
media: satellite, cable, terrestrial, etc. Each baseline system standard defined the channel coding and modulation
schemes for that transmission medium. The source coding was adapted from the MPEG-2 standard.
The design of these new systems has created a demand for a common understanding of measurement techniques and the
interpretation of measurement results.
The present document is an attempt to give recommendations in this field by defining a number of measurement
techniques in such detail that the results are actually comparable as long as the measurement is carried out in
compliance with the given definition.
Engineers seeking to apply the methods described in the present document should be familiar with the standards for the
respective baseline systems. Although most of the parameters specified in the present document are well known in
communications, most of them should be interpreted with respect to the new environment, especially the transmission
of digital TV signals or other related services.
The inclusion of each parameter in the present document is based on requirements from those who envisage having to
work alongside the defined procedures. This includes network operators and providers of equipment for network
installation, as well as manufacturers of Integrated Receiver Decoders (IRD) or test and measurement equipment.
The recommendations of the present document can be used:
- to set-up test beds or laboratory equipment for testing hardware for digital TV and other related services;
- to set these instruments to the appropriate parameters;
- to obtain unambiguous results that can be directly compared with results from other test set-ups;
- to form a potential basis for communicating results in an efficient way by using the definitions in the present
document as references.
They are not intended to describe a set of compulsory tests.
ETSI
15 ETSI TR 101 290 V1.2.1 (2001-05)
The recommendations are grouped in several clauses. Since the MPEG-2 TS is the signal format used for the inputs and
outputs of all baseline systems, clause 5 is devoted to the description of checking procedures for those parameters which
are accessible in the TS packet header, i.e. without decoding scrambled or encrypted data. The aim of these tests is the
provision of a simple and fast health check. It is meant neither as a MPEG-2 conformance test nor as a compliance test
for all DVB related issues.
Clause 6 contains the parameters which are commonly addressed by various transmission media. For example, the
measurement of the availability of transmission systems or links falls into this category, and it may be desirable to have
the same definition for availability independent of the actual system in use.
Clauses 7 and 8 address the parameters which are specific for cable and satellite, DVB-C and DVB-S, they are also
applicable to SMATV systems, DVB-CS, and possiblyMMDS systems such as DVB-MC and DVB-MS.
Clause 9 addressed parameters specific to the terrestrial DVB environment (DVB-T).
Clauses 6, 7, 8, and 9 of the present document follow the same structure. For each parameter there is a description of the
purpose of the recommended measurement procedure, the interface to which the measurement instrument should be
applied, and a description of the actual method of the measurement itself.
Apart from these clauses a number of annexes are included, containing recommendations for general aspects, examples
of test set-ups and certain requirements for the test and measurement equipment.
If the interfaces for a described measurement procedure are to be found within the transmitter, the notation is provided
in accordance with figures 4-1 and 9-1 for terrestrial. If the interfaces for the described measurement procedures are to
be found within the receiver (test receiver or IRD), the notation is provided in accordance with figures 4-2 and 9-2 for
terrestrial. These figures illustrate the general cases of a DVB transmitter and receiver, although certain functional
blocks only appear in certain systems.
Most of the parameters can be measured with standard equipment such as spectrum analysers or constellation analysers.
Other parameters are defined in a new way as a request to test and measurement equipment manufacturers to integrate
this functionality in their products.
Figure 4-1: Transmitter block diagram
Figure 4-2: Receiver block diagram
ETSI
16 ETSI TR 101 290 V1.2.1 (2001-05)
5 Measurement and analysis of the MPEG-2 Transport
Stream
5.1 General
The MPEG-2 Transport Stream (TS) is the specified input and output signal for all the baseline systems, i.e. for
satellite, cable, SMATV, MMDS/MVDS and terrestrial distribution, which are defined in the DVB world so far.
Therefore these interfaces are accessible in the transmission chain. Direct access is given on the transmitter side at the
input of the respective baseline system. At other interfaces where the signal occurs in modulated form, access is
possible by an appropriate demodulator that provides the TS interface as an output for further measurements.
5.2 List of parameters recommended for evaluation
The present document recommends in this clause a set of syntax and information consistency tests that can be applied to
an MPEG-2 TS at the parallel interface, or either of the serial interfaces defined in EN 50083-9 [10].
The following assumptions and guiding principles were used in developing these tests:
- the tests are mainly intended for continuous or periodic monitoring of MPEG-2 TSs in an operational
environment;
- these tests are primarily designed to check the integrity of a TS at source; clause 5.3 covers other aspects of TSs
in networks including impairments created by transport systems;
- the general aim of the tests is to provide a "health check" of the most important elements of the TS. The list of
the tests is not exhaustive;
- the tests are consistent with the MPEG-2 Conformance tests defined in ISO/IEC 13818-4 [2], they do not replace
them;
- the tests are consistent with the DVB-SI documents (EN 300 468 [7], TR 101 211 [8]), they do not replace them.
MPEG-2 and DVB-SI reserved values in the TS do not cause a test error indication.
In general the tests are performed on TS header information so that they are still valid when conditional access
algorithms are applied, however a few of the tests may only be valid for an unscrambled or descrambled TS.
The tests are not dependant on any decoder implementation for consistency of results. The MPEG-2 T-STD model
constraints, as defined in ISO/IEC 13818-1 [1] (MPEG-2 Systems), shall be satisfied as specified in
ISO/IEC 13818-4 [2] (MPEG-2 Compliance).
Off-line tests are performed under stable conditions, no discontinuity or dynamic change can occur during an off-line
test process.
Other digital performance parameters such as BER are not considered in this clause.
This clause tabulates the parameters which are recommended for continuous or periodic monitoring of the MPEG-2 TS.
The tests are grouped into three tables according to their importance for monitoring purposes.
The first table lists a basic set of parameters which are considered necessary to ensure that the TS can be decoded. The
second table lists additional parameters which are recommended for continuous monitoring. The third table lists
optional additional parameters which could be of interest for certain applications.
Any test equipment intended for the evaluation of these parameters should report test results by means of the indicators
itemized in the second column of the tables under exactly the preconditions described in the third column of the tables.
If an indicator is set, then the TS is in error. However, since the indicators do not cover the entire range of possible
errors, it cannot be concluded that there is no error if the indicator is not set.
ETSI
17 ETSI TR 101 290 V1.2.1 (2001-05)
If indicator 1.1 is activated then all other indicators are invalid. Each indicator is activated only as long as at least one
of the described preconditions is fulfilled.
NOTE: In the case of indicators requiring a minimum repetition rate of sections, it is intended that each and every
section that is present for this table should have the stated repetition rate.
5.2.1 First priority: necessary for de-codability (basic monitoring)
No. Indicator Precondition Reference
1.1 TS_sync_loss Loss of synchronization with consideration of
hysteresis parameters
ISO/IEC 13818-1 [1]:
clause 2.4.3.3 and annex G.01
1.2 Sync_byte_error Sync_byte not equal 0x47 ISO/IEC 13818-1 [1]:
clause 2.4.3.3
1.3 PAT_error PID 0x0000 does not occur at least every 0,5 s
a PID 0x0000 does not contain a table_id 0x00 ( i.e.
a PAT)
Scrambling_control_field is not 00 for PID 0x0000
ISO/IEC 13818-1 [1]:
clauses 2.4.4.3, 2.4.4.4
1.3.a
(note 1)
PAT_error_2 Sections with table_id 0x00 do not occur at least
every 0,5 s on PID 0x0000.
Section with table_id other than 0x00 found on PID
0x0000.
Scrambling_control_field is not 00 for PID 0x0000
TR 101 154 [4] 4.1.7
ISO/IEC 13818-1 [1]:
clauses 2.4.4.3, 2.4.4.4
1.4 Continuity_
count_error
Incorrect packet order
a packet occurs more than twice
lost packet
ISO/IEC 13818-1 [1]:
clauses 2.4.3.2, 2.4.3.3
1.5 PMT_error Sections with table_id 0x02, ( i. e. a PMT), do not
occur at least every 0,5 s on the PID which is
referred to in the PAT
Scrambling_control_field is not 00 for all PIDs
containing sections with table_id 0x02 (i.e. a PMT)
ISO/IEC 13818-1 [1]:
clauses 2.4.4.3, 2.4.4.4, 2.4.4.8
1.5.a
(note 2)
PMT_error_2 Sections with table_id 0x02, (i.e. a PMT), do not
occur at least every 0,5 s on each
program_map_PID which is referred to in the PAT
Scrambling_control_field is not 00 for all packets
containing information of sections with table_id
0x02 (i.e. a PMT) on each program_map_PID
which is referred to in the PAT
TR 101 154 [4] 4.1.7 (note 3)
ISO/IEC 13818-1 [1]:
clauses 2.4.4.3, 2.4.4.4, 2.4.4.8
1.6 PID_error Referred PID does not occur for a user specified
period.
ISO/IEC 13818-1 [1]:
clause 2.4.4.8
NOTE 1: Recommended for future implementations as a replacement of 1.3.
NOTE 2: Recommended for future implementations as a replacement of 1.5; this excludes specificly network_PIDs.
NOTE 3: In TR 101 154 [4], it is recommended that the interval between two sections should not exceed 100 ms.
For many applications it may be sufficient to check that the interval is not longer than 0.5 s.
TS_sync_loss
The most important function for the evaluation of data from the MPEG-2 TS is the sync acquisition. The actual
synchronization of the TS depends on the number of correct sync bytes necessary for the device to synchronize and on
the number of distorted sync bytes which the device can not cope with.
It is proposed that five consecutive correct sync bytes (ISO/IEC 13818-1 [1], clause G.01) should be sufficient for sync
acquisition, and two or more consecutive corrupted sync bytes should indicate sync loss.
After synchronization has been achieved the evaluation of the other parameters can be carried out.
Sync_byte_error
The indicator "Sync_byte_error" is set as soon as the correct sync byte (0x47) does not appear after 188 or 204 bytes.
This is fundamental because this structure is used throughout the channel encoder and decoder chains for
synchronization. It is also important that every sync byte is checked for correctness since the encoders may not
necessarily check the sync byte. Apparently some encoders use the sync byte flag signal on the parallel interface to
control randomizer re-seeding and byte inversion without checking that the corresponding byte is a valid sync byte.
ETSI
18 ETSI TR 101 290 V1.2.1 (2001-05)
PAT_error
The Program Association Table (PAT), which only appears in PID 0x0000 packets, tells the decoder what programs are
in the TS and points to the Program Map Tables (PMT) which in turn point to the component video, audio and data
streams that make up the program (figure 5-2).
If the PAT is missing then the decoder can do nothing, no program is decodable.
Nothing other than a PAT should be contained in a PID 0x0000.
PAT_error_2
The reworded description of the error in PAT_error_2 refers to the possibility that the Program Association Table may
consist of several (consecutive) sections with the same table_id 0x00.
Continuity_count_error
For this indicator three checks are combined. The preconditions "Incorrect packet order" and "Lost packet" could cause
problems for IRD which are not equipped with additional buffer storage and intelligence. It is not necessary for the test
equipment to distinguish between these two preconditions as they are logically OR-ed, together with the third
precondition, into one indicator.
The latter is also covering the packet loss that may occur on ATM links, where one lost ATM packet would cause the
loss of a complete MPEG-2 packet.
The precondition "a packet occurs more than twice" may be symptomatic of a deeper problem that the service provider
would like to keep under observation.
PMT_error
The Program Association Table (PAT) tells the decoder how many programs there are in the stream and points to the
PMTs which contain the information where the parts for any given event can be found. Parts in this context are the
video stream (normally one) and the audio streams and the data stream (e.g. Teletext). Without a PMT the
corresponding program is not decodable.
PID_error
It is checked whether there exists a data stream for each PID that occurs. This error might occur where TS are
multiplexed, or demultiplexed and again remultiplexed.
The user specified period should not exceed 5 s for video or audio PIDs (see note). Data services and audio services
with ISO 639 [17] language descriptor with type greater than '0' should be excluded from this 5 s limit.
NOTE: For PIDs carrying other information such as sub-titles, data services or audio services with ISO 639 [17]
language descriptor with type greater than '0', the time between two consecutive packets of the same PID
may be significantly longer.
In principle, a different user specified period could be defined for each PID.
ETSI
19 ETSI TR 101 290 V1.2.1 (2001-05)
5.2.2 Second priority: recommended for continuous or periodic monitoring
No. Indicator Precondition Reference
2.1 Transport_error Transport_error_indicator in the TS-Header is set to
"1"
ISO/IEC 13818-1 [1]:
clauses 2.4.3.2, 2.4.3.3
2.2 CRC_error CRC error occurred in CAT, PAT, PMT, NIT, EIT,
BAT, SDT or TOT table
ISO/IEC 13818-1 [1]:
clauses 2.4.4, annex B
EN 300 468 [7]: clause 5.2
2.3 PCR_error (note) PCR discontinuity of more than 100 ms occurring
without specific indication.
Time interval between two consecutive PCR values
more than 40 ms
ISO/IEC 13818-1 [1]:
clauses 2.4.3.4, 2.4.3.5
ISO/IEC 13818-4 [2]:
clause 9.11.3
TR 101 154 [4]: clause 4.5.4
2.3a PCR_repetition_
error
Time interval between two consecutive PCR values
more than 40 ms
TR 101 154 [4]: clause 4.1.5.3
2.3b PCR_discontinuity_i
ndicator_error
The difference between two consecutive PCR
values (PCRi+1 – PCRi) is outside the range of
0...100 ms without the discontinuity_ indicator set
ISO/IEC 13818-1 [1]:
clauses 2.4.3.4, 2.4.3.5
ISO/IEC 13818-4 [2]:
clause 9.1.1.3
2.4 PCR_accuracy_
error
PCR accuracy of selected programme is not within
±500 ns
ISO/IEC 13818-1 [1]:
clause 2.4.2.2
2.5 PTS_error PTS repetition period more than 700 ms ISO/IEC 13818-1 [1]:
clauses 2.4.3.6, 2.4.3.7, 2.7.4
2.6 CAT_error Packets with transport_scrambling_control not 00
present, but no section with table_id = 0x01 (i.e. a
CAT) present
Section with table_id other than 0x01
(i.e. not a CAT) found on PID 0x0001
ISO/IEC 13818-1 [1]:
clause 2.4.4
NOTE: The old version of PCR_error (2.3) is a combination of the more specific errors PCR_repetition_error
(2.3.a) and PCR_discontinuity_indicator_error (2.3.b) by a logical 'or' function. It is kept in the present
document for reasons of consistency of existing implementations. For new implementations it is
recommended that the indicators 2.3.a and 2.3.b are used only.
Transport_error
The primary Transport_error indicator is Boolean, but there should also be a resettable binary counter which counts the
erroneous TS packets. This counter is intended for statistical evaluation of the errors. If an error occurs, no further error
indication should be derived from the erroneous packet.
There may be value in providing a more detailed breakdown of the erroneous packets, for example, by providing a
separate Transport_error counter for each program stream or by including the PID of each erroneous packet in a log of
Transport_error events. Such extra analysis is regarded as optional and not part of this recommendation.
CRC_error
The CRC check for the CAT, PAT, PMT, NIT, EIT, BAT, SDT and TOT indicates whether the content of the
corresponding table is corrupted. In this case no further error indication should be derived from the content of the
corresponding table.
PCR_error
The PCRs are used to re-generate the local 27 MHz system clock. If the PCR do not arrive with sufficient regularity
then this clock may jitter or drift. The receiver/decoder may even go out of lock. In DVB a repetition period of not more
than 40 ms is recommended.
PCR_repetition_error
The PCRs are used to re-generate the local 27 MHz system clock. If the PCR do not arrive with sufficient regularity
then this clock may jitter or drift. The receiver/decoder may even go out of lock. In DVB a repetition period of not more
than 40 ms is recommended.
The error indication that may result from the check of this repetition period should be called PCR_repetition_error in
future implementations (after the release of the present document).
PCR_discontinuity_indicator_error
The PCR_discontinuity_indicator_error is set in the case that a discontinuity of the PCR values occurs that has not been
signalled appropriately by the discontinuity indicator. The usage of this indicator is recommended for future
implementations (after the release of the present document).
ETSI
20 ETSI TR 101 290 V1.2.1 (2001-05)
PCR_accuracy_error
The accuracy of ±500 ns is intended to be sufficient for the colour subcarrier to be synthesized from system clock.
This test should only be performed on a constant bitrate TS as defined in ISO/IEC 13818-1 [1] clause 2.1.7.
Further information on PCR jitter measurements is given in clause 5.3.2. and annex I.
PTS_error
The Presentation Time Stamps (PTS) should occur at least every 700 ms. They are only accessible if the TS is not
scrambled.
CAT_error
The CAT is the pointer to enable the IRD to find the EMMs associated with the CA system(s) that it uses. If the CAT is
not present, the receiver is not able to receive management messages.
5.2.3 Third priority: application dependant monitoring
No. Indicator Precondition Reference
3.1 NIT_error (note 2) Section with table_id other than 0x40 or 0x41 or
0x72 (i. e. not an NIT or ST) found on PID 0x0010
No section with table_id 0x40 or 0x41 (i.e. an NIT)
in PID value 0x0010 for more than 10 s
EN 300 468 [7]: clause 5.2.1
TR 101 211 [8]:
clauses 4.1, 4.4
3.1.a NIT_actual_error Section with table_id other than 0x40 or 0x41 or
0x72 (i. e. not an NIT or ST) found on PID 0x0010
No section with table_id 0x40 (i.e. an NIT_actual)
in PID value 0x0010 for more than 10 s.
Any two sections with table_id = 0x40 (NIT_actual)
occur on PID 0x0010 within a specified value (25
ms or lower).
EN 300 468 [7]: clause 5.2.1,
5.1.4
TR 101 211 [8]:
clauses 4.1, 4.4,
3.1.b NIT_other_error Interval between sections with the same
section_number and table_id = 0x41 (NIT_other)
on PID 0x0010 longer than a specified value (10s
or higher).
TR 101 211 [8] clause 4.4.
3.2 SI_repetition_
error
Repetition rate of SI tables outside of specified
limits.
EN 300 468 [7]: clause 5.1.4
TR 101 211 [8]: clause 4.4
3.3 Buffer_error TB_buffering_error
overflow of transport buffer (TBn)
TBsys_buffering_error
overflow of transport buffer for system information
(Tbsys)
MB_buffering_error
overflow of multiplexing buffer (MBn) or
if the vbv_delay method is used:
underflow of multiplexing buffer (Mbn)
EB_buffering_error
overflow of elementary stream buffer (EBn) or
if the leak method is used:
underflow of elementary stream buffer (EBn)
though low_delay_flag and DSM_trick_mode_flag
are set to 0
else (vbv_delay method)
underflow of elementary stream buffer (EBn)
B_buffering_error
overflow or underflow of main buffer (Bn)
Bsys_buffering_error
overflow of PSI input buffer (Bsys)
ISO/IEC 13818-1 [1]:
clause 2.4.2.3
ISO/IEC 13818-4 [2]:
clauses 9.11.2, 9.1.4
3.4 Unreferenced_PID PID (other than PAT, CAT, CAT_PIDs, PMT_PIDs,
NIT_PID, SDT_PID, TDT_PID, EIT_PID,
RST_PID, reserved_for_future_use PIDs, or PIDs
user defined as private data streams) not referred
to by a PMT within 0,5 s (note 1).
EN 300 468 [7]: clause 5.1.3
ETSI
21 ETSI TR 101 290 V1.2.1 (2001-05)
No. Indicator Precondition Reference
3.4.a Unreferenced_PID PID (other than PMT_PIDs, PIDs with numbers
between 0x00 and 0x1F or PIDs user defined as
private data streams) not referred to by a PMT or a
CAT within 0,5 s
EN 300 468 [7]: clause 5.1.3
3.5 SDT_error (note 3) Sections with table_id = 0x42 (SDT, actual TS) not
present on PID 0x0011 for more than 2 s
Sections with table_ids other than 0x42, 0x46,
0x4A or 0x72 found on PID 0x0011
EN 300 468 [7]: clause 5.1.3
TR 101 211 [8]:
clauses 4.1, 4.4
3.5.a SDT_actual_error Sections with table_id = 0x42 (SDT, actual TS) not
present on PID 0x0011 for more than 2 s
Sections with table_ids other than 0x42, 0x46,
0x4A or 0x72 found on PID 0x0011.
Any two sections with table_id = 0x42
(SDT_actual) occur on PID 0x0011 within a
specified value (25 ms or lower).
EN 300 468 [7]: clause 5.2.3,
5.1.4
TR 101 211 [8]:
clauses 4.1, 4.4
3.5.b SDT_other_error Interval between sections with the same
section_number and table_id = 0x46 (SDT, other
TS) on PID 0x0011 longer than a specified value
(10s or higher).
TR 101 211 [8] clause 4.4
3.6 EIT_error (note 4) Sections with table_id = 0x4E (EIT-P/F,
actual TS) not present on PID 0x0012 for more
than 2 s
Sections with table_ids other than in the range
0x4E - 0x6F or 0x72 found on PID 0x0012
EN 300 468 [7]: clause 5.1.3
TR 101 211 [8]:
clauses 4.1, 4.4
3.6.a EIT_actual_error Section '0' with table_id = 0x4E (EIT-P,
actual TS) not present on PID 0x0012 for more
than 2 s
Section '1' with table_id = 0x4E (EIT-F,
actual TS) not present on PID 0x0012 for more
than 2 s
Sections with table_ids other than in the range
0x4E - 0x6F or 0x72 found on PID 0x0012.
Any two sections with table_id = 0x4E (EIT-P/F,
actual TS) occur on PID 0x0012 within a specified
value (25ms or lower).
EN 300 468 [7]: clause 5.2.4,
5.1.4
TR 101 211 [8]:
clauses 4.1, 4.4
3.6.b EIT_other_error Interval between sections '0' with table_id = 0x4F
(EIT-P, other TS) on PID 0x0012 longer than a
specified value (10s or higher);
Interval between sections '1' with table_id = 0x4F
(EIT-F, other TS) on PID 0x0012 longer than a
specified value (10s or higher).
TR 101 211 [8] clause 4.4
3.6.c EIT_PF_error If either section ('0' or '1') of each EIT P/F subtable
is present both must exist. Otherwise
EIT_PF_error should be indicated
EN 300 468 [7] caluse 5.2.4.
3.7 RST_error Sections with table_id other than 0x71 or 0x72
found on PID 0x0013.
Any two sections with table_id = 0x71 (RST) occur
on PID 0x0013 within a specified value (25 ms or
lower).
EN 300 468 [7]: clause 5.1.3
3.8 TDT_error Sections with table_id = 0x70 (TDT) not present
on PID 0x0014 for more than 30 s
Sections with table_id other than 0x70, 0x72 (ST)
or 0x73 (TOT) found on PID 0x0014.
Any two sections with table_id = 0x70 (TDT) occur
on PID 0x0014 within a specified value (25 ms or
lower).
EN 300 468 [7]: clauses 5.1.3,
5.2.6
TR 101 211 [8]:
clauses 4.1, 4.4
3.9 Empty_buffer_error Transport buffer (TBn) not empty at least once per
second
or
transport buffer for system information (TBsys) not
empty at least once per second
or
if the leak method is used
multiplexing buffer (MBn) not empty at least once
per second.
ISO/IEC 13818-1 [1]:
clauses 2.4.2.3, 2.4.2.6
ISO/IEC 13818-9 [3]:
annex E
ISO/IEC 13818-4 [2]:
clauses 9.1.1.2, 9.1.4
ETSI
22 ETSI TR 101 290 V1.2.1 (2001-05)
No. Indicator Precondition Reference
3.10 Data_delay_error Delay of data (except still picture video data)
through the TSTD buffers superior to 1 second;
or
delay of still picture video data through the TSTD
buffers superior to 60 s.
ISO/IEC 13818-1 [1]:
clauses 2.4.2.3, 2.4.2.6
NOTE 1: It is assumed that transition states are limited to 0,5 s, and these transitions should not cause error
indications.
NOTE 2: The old version of NIT_error (3.1) has been split into the more specific errors NIT_actual_error (3.1.a)
and NIT_other_error (3.1.b). The old version is kept in the document for reasons of consistency of
existing implementations. For new implementations it is recommended that the indicators 3.1.a and
3.1.b are used only.
NOTE 3: The old version of SDT_error (3.5) has been split into the more specific errors SDT_actual_error (3.5.a)
and SDT_other_error (3.5.b). The old version is kept in the present document for reasons of
consistency of existing implementations. For new implementations it is recommended that the
indicators 3.5.a and 3.5.b are used only.
NOTE 4: The old version of EIT_error (3.6) has been split into the more specific errors EIT_actual_error (3.6.a),
EIT_other_error (3.6.b) and EIT_PF_error (3.6.c). The old version is kept in the present document for
reasons of consistency of existing implementations. For new implementations it is recommended that
the indicators 3.6.a, 3.6.b and 3.6.c are used only.
NIT_error
Network Information Tables (NITs) as defined by DVB contain information on frequency, code rates, modulation,
polarization etc. of various programs which the decoder can use. It is checked whether NITs are present in the TS and
whether they have the correct PID.
NIT_actual_error
Network Information Tables (NITs) as defined by DVB contain information on frequency, code rates, modulation,
polarization etc. of various programs which the decoder can use. It is checked whether the NIT related to the respective
TS is present in this TS and whether it has the correct PID.
NIT_other_error
Further Network Information Tables (NITs) can be present under a separate PID and refer to other TSs to provide more
information on programmes available on other channels. Their distribution is not mandatory and the checks should only
be performed if they are present.
SI_repetition_error
For SI tables a maximum and minimum periodicity are specified in EN 300 468 [7] and TR 101 211 [8] . This is
checked for this indicator. This indicator should be set in addition to other indicators of repetition errors for specific
tables.
Buffer_error
For this indicator a number of buffers of the MPEG-2 reference decoder are checked whether they would have an
underflow or an overflow.
Unreferenced_PID
Each non-private program data stream should have its PID listed in the PMTs.
SDT_error
The SDT describes the services available to the viewer. It is split into sub-tables containing details of the contents of the
current TS (mandatory) and other TS (optional). Without the SDT, the IRD is unable to give the viewer a list of what
services are available. It is also possible to transmit a BAT on the same PID, which groups services into "bouquets".
SDT_actual_error
The SDT (Service Description Table) describes the services available to the viewer. It is split into sub-tables containing
details of the contents of the current TS (mandatory) and other TS (optional). Without the SDT, the IRD is unable to
give the viewer a list of what services are available. It is also possible to transmit a BAT on the same PID, which groups
services into "bouquets".
SDT_other_error
This check is only performed if the presence of a SDT for other TSs has been established.
ETSI
23 ETSI TR 101 290 V1.2.1 (2001-05)
EIT_error
The EIT (Event Information Table) describes what is on now and next on each service, and optionally details the
complete programming schedule. The EIT is divided into several sub-tables, with only the "present and following"
information for the current TS being mandatory. The EIT schedule information is only accessible if the TS is not
scrambled.
EIT_actual_error
The EIT (Event Information Table) describes what is on now and next on each service, and optionally details the
complete programming schedule. The EIT is divided into several sub-tables, with only the "present and following"
information for the current TS being mandatory. If there are no 'Present' or 'Following' events, empty EIT sections will
be transmitted according to TR 101 211 [8]. The EIT schedule information is only accessible if the TS is not scrambled.
EIT_other_error
This check is only performed if the presence of an EIT for other TSs has been established.
RST_error
The RST is a quick updating mechanism for the status information carried in the EIT.
TDT_error
The TDT carries the current UTC time and date information. In addition to the TDT, a TOT can be transmitted which
gives information about a local time offset in a given area.
The carriage of the following tables:
- NIT_other;
- SDT_other;
- EIT_P/F_other;
- EIT_schedule_other;
- EIT_schedule_actual,
is optional and therefore these tests should only be performed when the respective table is present.
When these tables are present this will be done automatically by measuring the interval rather than the occurrence of the
first section.
As a further extension of the checks and measurements mentioned above an additional test concerning the SI is
recommended: all mandatory descriptors in the SI tables should be present and the information in the tables should be
consistent.
ETSI
24 ETSI TR 101 290 V1.2.1 (2001-05)
Figure 5-1: Indicators related to TS syntax
ETSI
25 ETSI TR 101 290 V1.2.1 (2001-05)
Figure 5-2: Indicators related to TS structure
5.3 Measurement of MPEG-2 Transport Streams in networks
5.3.1 Introduction
A MPEG-2 Transport Stream that is transmitted over any real network, is exposed to certain effects caused by the
network components which are not ideally transparent. One of the pre-dominant effects is the acquisition of jitter in
relation to the PCR values and their position in the TS. The parameters defined in 5.3.2 describe the various jitter
components which can be differentiated by demarcation frequencies.
For the measurement of bitrates of Transport Streams, the requirements vary significantly for constant bitrate TS and
partial TS/ variable bitrate TS. The application of statistical multiplexers led to more dynamic variations in the bitrate,
especially of the video components. Other services such as opportunistic data transmission, have typical features which
again differ in terms of occurence or presence of the service and the variation of bitrates. In 5.3.3 several profiles are
defined to accommodate the majority of such applications, and which can be applied for monitoring and localization of
failures.
5.3.2 System clock and PCR measurements
5.3.2.1 Reference model for system clock and PCR measurements
This clause presents a reference model for any source of a transport stream (TS) concerning the generation of PCR
values and delivery delays. It models all the timing effects visible at the TS interface point. It is not intended to
represent all the mechanisms by which these timing effects could arise in real systems.
ETSI
26 ETSI TR 101 290 V1.2.1 (2001-05)
Reference clock
f = 27MHz + fdev(t)
PCR counter
PCR inaccuracy source
Mp,i
Np,i
+
D +Ji
Delivery timing
delay
A
B C
Figure 5-3: Reference model
Reference points are indicated by dashed lines. This is a model of an encoder/multiplexer (up to reference point B) and
a physical delivery mechanism or communications network (between reference points B and C). The components of the
model to the left of reference point B are specific to a single PCR PID. The components of the model to the right of
reference point B relate to the whole Transport Stream. Measuring equipment can usually only access the TS at
reference point C.
The model consists of a system clock frequency oscillator with a nominal frequency of 27 MHz, but whose actual
frequency deviates from this by a function fdev(p, t). This function depends on the time (t) and is specific to a single
PCR PID (p). The "Frequency Offset PCR_FO" measures the value of fdev(p, t). The "Drift Rate PCR_DR" is the rate
of change with time of fdev(p, t).
The system clock frequency oscillator drives a PCR counter which generates an idealized PCR count, Np,i. p refers to
the specific PCR PID p and i refers to the bit position in the transport stream. To this is added a value from a PCR
inaccuracy source, Mp,i to create the PCR value seen in the stream, Pp,i. The simple relationship between these values
is:
Pp,i = N p,i + M p,i
Equation 1
Mp,i represents the "Accuracy PCR_AC".
The physical delivery mechanism or communications network beyond point B introduces a variable delay between the
departure time Ti and the arrival time Ui of bits:
i i i U − T = D + J
Equation 2
In the case of a PCR, Ui is the time of arrival of the last bit of the last byte containing the PCR base (ISO/IEC13818-1
[1], clause 2.4.3.5). D is a constant representing the mean delay through the communications network. Ji represents the
jitter in the network delay and its mean value over all time is defined to be zero. Ji+ Mp,i is measured as the "Overall
Jitter PCR_OJ".
In the common case where the the Transport Stream is constant bitrate, at reference point B the Transport Stream is
being transmitted at a constant bitrate Rnom. It is important to note that in this reference model this bitrate is accurate
and constant; there is no error contribution from varying bitrate. This gives us an additional equation for the departure
time of packets:
ETSI
27 ETSI TR 101 290 V1.2.1 (2001-05)
nom
i R
i
T = T + 0
Equation 3
T0 is a constant representing the time of departure of the zero'th bit. Combining equations 2 and 3 we have for the
arrival time:
i
nom
i D J
R
i
U = T + + + 0
Equation 4
5.3.2.2 Measurement descriptions
The following measurements require a demarcation frequency for delimiting the range of drift rate and jitter frequencies
of the timing variations of PCRs and/ or TSs.
The demarcation frequency used should be chosen from the following table and indicated with the measurement results.
In clause I.5 a description can be found for the derivation of the demarcation frequencies.
ETSI
28 ETSI TR 101 290 V1.2.1 (2001-05)
Table 5.1: Profiles for jitter and drift rate measurements
Profile Demarcation
frequency
Comments
MGF1 10 mHz This profile is provided to give the total coverage of frequency components
included in the timing impairments of PCR related measurements.
This profile provides the most accurate results in accordance with the limits
specified in ISO/IEC 13818-1 [1], clause 2.4.2.1. If jitter or drift rate
measurements are found out of specification when using other profiles, it is
suggested to use this one for better accuracy.
MGF2 100 mHz This profile is accounting for intermediate benefits between the profiles MGF1
and MGF3, by giving reasonable measurement response as well as reasonable
account for low frequency components of the timing impairments.
MGF3 1 Hz This profile provides faster measurement response by taking in account only the
highest frequency components of the timing impairments. This profile is expected
to be sufficient in many applications.
MGF4 Manufacturer
defined
This profile will provide any benefit that the manufacturer may consider as useful
when it is designed and implemented in a measurement instrument. The
demarcation frequency has to be supplied with the measurement result.
Optionally any other data that the manufacturer may consider to be relevant may
be supplied.
For testing against ISO/IEC13818-9 [3] (±25 μs jitter limit) a demarcation
frequency of 2 mHz is required. A filter for such demarcation may be
implemented under this MGF4 profile.
5.3.2.3 Program Clock Reference - Frequency Offset PCR_FO
Definition PCR_FO is defined as the difference between the program clock frequency and the nominal clock
frequency (measured against a reference which is not PCR derived, neither TS derived).
The units for the parameter PCR_FO should be in Hz according to:
Measured Frequency - Nominal Frequency,
or in ppm expressed as:
[Measured Frequency (in Hz) – Nominal Frequency(in Hz)]/Nominal Frequency (in MHz).
Purpose The original frequency of the clock used in the digital video format before compression (program clock) is
transmitted to the final receiver in form of numerical values in the PCR fields. The tolerance as specified
by ISO/IEC 13818-1 [1] is ±810 Hz or ±30 ppm.
Interface For example at Interface G in figure I-8 of annex I.
Method Refer to annex I for a description of a measurement method.
5.3.2.4 Program Clock Reference – Drift Rate PCR_DR
Definition PCR_DR is defined as the first derivative of the frequency and is measured on the low frequency
components of the difference between the program clock frequency and the nominal clock frequency
(measured against a reference which is not PCR derived, neither TS derived).
The format of the parameter PCR_DR should be in mHz/s (@ 27 MHz) or ppm/ hour.
Purpose The measurement is designed to verify that the frequency drift, if any, of the program clock frequency is
below the limits set by ISO/IEC 13818-1 [1]. This limit is effective only for the low frequency
components of the variations as indicated by the demarcation frequency described in annex I.
The tolerance as specified by ISO/IEC 13818-1 [1] is ±75 mHz/s@ 27 MHz or ±10 ppm/ hour.
Interface For example at Interface H in figure I-8 of annex I.
Method Refer to annex I for a description of a measurement method.
ETSI
29 ETSI TR 101 290 V1.2.1 (2001-05)
5.3.2.5 Program Clock Reference - Overall Jitter PCR_OJ
Definition PCR_OJ is defined as the instantaneous measurement of the high frequency components of the difference
between when a PCR should have arrived at a measurement point (based upon previous PCR values, its
own value and a reference which is not PCR or TS derived) and when it did arrive.
The format of the parameter PCR_OJ should be in nanoseconds.
Purpose The PCR_OJ measurement is designed to account for all cumulative errors affecting the PCR values
during program stream generation, multiplexing, transmission, etc. All these effects appear as jitter at the
receiver but they are a combination of PCR inaccuracies and jitter in the transmission. This value can be
compared against the maximum error specification by ISO/IEC 13818-1 [1] for PCR Accuracy of ±500 ns
only if the jitter in the transmission is assumed to be zero.
Interface For example at Interface J in figure I-8 of annex I.
Method Refer to annex I for a description of a measurement method.
5.3.2.6 Program Clock Reference – Accuracy PCR_AC
Definition The accuracy of the PCR values PCR_AC is defined as the difference between the actual PCR value and
the value it should have in the TS represented by the byte index for its actual position. This can be
calculated for constant bitrate TS, the measurement may NOT produce meaningful results in variable
bitrate TS.
The units for the parameter PCR_AC should be in nanoseconds.
Purpose This measurement is designed to indicate the total error included in the PCR value with respect to its
position in the TS.
The tolerance as specified by ISO/IEC 13818-1 [1] is ±500 ns.
This measurement is considered to be valid for both: real time and off-line measurements.
The measurement should trigger the indicator under paragraph 5.2.2. item 2.4.
Interface For example at Interface E in figure I-6 of annex I.
Method Refer to annex I for a description of a measurement method.
NOTE: Note that PCR Accuracy is defined by ISO/IEC 13818-1 [1]: "A tolerance is specified for the PCR values.
The PCR tolerance is defined as the maximum inaccuracy allowed received PCRs. This inaccuracy may
be due to imprecision in the PCR values or to PCR modification during re-multiplexing. It does not
include errors in packet arrival time due to network jitter or other causes".
5.3.3 Bitrate measurement
The bitrate value from a measurement system depends on a number of parameters:
- when the bitrate measurement is started;
- what is counted (packets, bytes, bits);
- the time duration (gate) over which the bitrate is measured;
- the way in which the time-gate function moves between measurements (timeSlice).
ETSI
30 ETSI TR 101 290 V1.2.1 (2001-05)
5.3.3.1 Bitrate measurement algorithm
This clause defines the parameter "MG bitrate" which is an instantaneous bitrate value. The bitrate is averaged over a
fixed time gate (or "window"). This gating function is moved by a discrete time slice (or interval) to produce the bitrate
value for each time slice. (The window "hops" from one time slice to the next) The items that are counted can be bits,
bytes or Transport Stream packets, and the meaning of the measured value should be made clear by accurate labelling
(see Nomenclature below). The measurement can be applied to the entire Transport Stream or a partial transport stream
obtained by applying a PID filter or even a filter to remove packet headers.
The following equation defines "MG bitrate":
= −
= − ×
Τ
=
1
0
_ _ _ _ _ _
τ
τ
n N
t n t n num elements in timeSlice
elementSize
MG bitrate at timeSlice
Where:
N is the integer number of time slices during the time gate.
T = Nτ is the duration of the time gate in seconds.
τ is the width of each time slice in seconds.
element is the fundamental unit which is being counted by the bitrate measurement algorithm.
elementSize is the size (measured in the appropriate units) of the element being measured. For example if
bitrate units are packets/s then the elementSize must be expressed in packets. If bitrate units are
bits/s then the elementSize is expressed in bits. Hence if an element ia a 188 byte packets then we
can express elementSize as:
elementSize = 188 bytes/packet × 8 bits/byte = 1 504 bits
num_elements_in_timeSlice is the integer number of element starts which have occurred in the timeSlice. If an
element is a 188 byte packet then this corresponds to counting sync bytes. If an
element is a byte then this may correspond to counting the first bit in transmission
order on a serial link.
The units of MG_bitrate_at_timeSlicet are not part of this specification, but must be the same as the units used to
express elementSize. This is because the bitrate can be expressed in a number of different ways as is described in the
Nomenclature clause below.
The measurement is discrete. A new measurement value is available every timeSlice and is held for the duration of a
timeSlice. Display of a bitrate value in a piece of measurement equipment may not be a precise display of this value as
is indicated in figure 5-4.
measure Process
for display
Process
for alarms
Process
for statistics
etc …
link to display
Remote
display
Interface
MGi1
Interface
MGi2
Figure 5-4: Display of a bitrate value
ETSI
31 ETSI TR 101 290 V1.2.1 (2001-05)
5.3.3.2 Preferred values for Bitrate Measurement
The preferred values for the algorithm are application dependent. One set of values may be appropriate for monitoring
and another may be appropriate for precise measurements. In order to have consistent measurements between different
equipment vendors, the following profiles are defined. (Note that the timeSlice interval τ can be expressed as a time or
as a frequency for precision).
MG
Profile
Profile Description Stream Type/Rate τ N T=Nτ element
MGB1 This Profile is best geared towards
applications where the bitrate is
constant or slowly varying. It is
compatible with much equipment
developed before this specification
was created.
All 1 s 1 1 s 188 byte
packet
MGB2 This Profile provides overall consistent
rate calculations while providing
reasonable accuracy for most
monitoring and troubleshooting
applications. It is inteneded for CBR
measurements whereas rapidly
varying bitrates are more appropriately
measured with the MGB3 or MGB4
profiles.
All 100 ms 10 1 s 188 byte
packet
MGB3 This Profile provides for tracking of
small variations in the multiplex rate of
each element.
All 1/90
kHz
1 800 20 ms 188 byte
packet
MGB4 This Profile provides for a longer term
average for rate calculation but with
repeatability between two different
measurements of the same data.
All 1/90
kHz
9×104 1 s 188 byte
packet
MGB5 This Profile allows the user to tune
bitrate calculations based on the
parameters that are most appropriate
for a particluar transport stream.
It is very important that when this is
done, the nomenclature used to define
the bitrate clearly shows that bitrates
for components are not directly
comparable with each other:
TS@MGB1
video@MGB3
audio@MGB4
the_rest@188,1s,100s
etc.
This follows the nomenclature guide in
this specification and shows that it is
unlikely that the sum of the bitrates of
the TS components will equal the
overall transport stream rate.
Complete or partial
transport stream
User
Def.
User
Def.
User
Def.
188 byte
packet
Applications of the profiles are given in the informative annex J.
5.3.3.3 Nomenclature
It is important to display bitrate values in a way which allows comparison. Correct nomenclature can indicate for
example that correction factors need to be applied to convert from a 204 byte packet bitrate measurement to a 188 byte
packet measurement. This recommendation is for the "MG-bitrate" nomenclature. If the "MG bitrate" algorithm has
been used, then bitrates are of the form:
<bitrate_value> <units>@ MGprofile
or <bitrate_value> <units> @ MG<element>, <timeslice>, <time_gate> [,<filter>]
ETSI
32 ETSI TR 101 290 V1.2.1 (2001-05)
For example if the full transport stream bitrate of a 204 byte packet system is to be measured, then it is important to
know the size of the packet (i.e. the elementSize) and the size of the time window which was measured to ensure
repeatability. Hence a bitrate should be expressed as:
10,300 Mbit/s@ MG 204,1/90 kHz,1,1s example 1
It is assumed by default that the bitrate was for the full transport stream.
If the bitrate of all the service components for a service called "Test Transmission" (i.e. all PIDs listed in the PMT + the
bitrate of the PMT excluding the bitrate of EITp and EITf for that service) is to be measured, then it would be expressed
as:
4,154 Mbit/s@ MG 188, 1/90 kHz,1s,service: Test Transmission example 2
or 4,154 Mbit/s @ MGB4, service: Test Transmission
To express example 2 as a percentage of the total bitrate in example 1, it is obvious now that a 188/204 correction factor
needs to be applied before the division takes place:
Test Transmission = 100 × (4,154 × 204/188 )/10,300 % of bitrate
43,8 % of bitrate
Note that this nomenclature is independent of the measurement technique, but is vital to allow results to be compared.
Note also that when writing MG-bitrate measurements, the values kbit/s and Mbit/s are taken to mean 103 bits per
second and 106 bits per second respectively. It is also recommended that the values kB/s (103 bytes/s) and MB/s
(106 bytes/s) are not used.
5.3.4 Consistency of information check
The information provided in the various SI/ PSI tables in different Transport Streams needs to be consistent and
coherent to provide access to all services for the user. Wherever these tables are created, modified or extracted, there is
a need for checking the tables of the outgoing Transport Stream.
In many cases, these applications are user-defined in the sense that providers and operators may wish to minimize the
complexity of these checks.
As a first example for such a check, the Transport_Stream_ID check is defined hereafter.
5.3.4.1 Transport_Stream_ID check
Definition Each MPEG-2 Transport Stream should be identifiable by its Transport_Stream_ID carried in the PAT.
Purpose As DVB networks become more and more complex, there is an increased risk of transmitting the wrong
Transport Stream. Providers and operators may wish to make sure that the TS they actually process is the
intended one.
Interface A, Z
Method The Transport Stream ID (as referenced in the PAT) should be checked and the actual TS ID should be
compared with a user defined value. By this it can be tested whether the actual Transport Stream is the
correct one.
5.3.5 TS parameters in transmission systems with reduced SI data
Certain transmission systems, e.g. DSNG Transport Streams conforming to EN 301 210 [18] contain simplified PSI/SI
information (see annex D of EN 301 210 [18]). When testing such Transport Streams, the following tables indicate
which of the tests recommended in clause 5.2 can be used.
ETSI
33 ETSI TR 101 290 V1.2.1 (2001-05)
No. Indicator Comment
1.1 TS_sync_loss Essential for access to TS data
1.2 Sync_byte_error May not necessarily prevent
decoding of content
1.3 PAT_error Essential for access to TS data
1.3.a PAT_error_2 Essential for access to TS data
1.4 Continuity_ count_error May not necessarily prevent
decoding of content
1.5 PMT_error Essential for access to TS data
1.5.a PMT_error_2 Essential for access to TS data
1.6 PID_error May not necessarily prevent
decoding of content
No. Indicator Comment
2.1 Transport_error
2.2 CRC_error Applies to PAT and PMT only
2.3 PCR_error
2.3a PCR_repetition_error
2.3b PCR_discontinuity_indicator_error
2.4 PCR_accuracy_error
2.5 PTS_error
2.6 CAT_error
No. Indicator Comment
3.3 Buffer_error
3.4 Unreferenced_PID
3.4.a Unreferenced_PID
3.9 Empty_buffer_error
3.10 Data_delay_error
5.4 Measurement of availability at MPEG-2 Transport Stream
level
Definitions of error events
The following definitions are used to establish criteria for System Availability, Link Availability, and System Error
Performance (e.g. for coverage measurement purposes) for distribution networks such as satellite (DVB-S and
DVB-DSNG), cable (DVB-C), terrestrial (DVB-T) and microwave systems (DVB-MS, DVB-MC and DVB-MT) as
well as for contribution networks (DVB-PDH ETS 300 813 [19] and DVB-SDH ETS 300 814 [20]).
These definitions may also be used to test the performance of TSs in IRDs via Common Interfaces.
ETSI
34 ETSI TR 101 290 V1.2.1 (2001-05)
Table 5.2: Error Events
5.4.1 Severely Disturbed Period
(SDP):
A period of sync loss (as defined in clause 5.2.1 of the present
document, parameter 1.1) or loss of signal.
5.4.2 Errored Block (EB): An MPEG-2 TS packet with one or more uncorrectable errors, which
is indicated by the transport_error_indicator flag set.
See clause 5.2.2.
5.4.3 Errored Time Interval (ETI): A given time interval with one or more EBs.
5.4.3.a Errored Second (ES): A specific case of the ETI where the given time interval is one
second.
5.4.4 Severely Errored Time Interval
(SETI):
A given time interval which contains greater than a specified
percentage of errored blocks, or at least one SDP or part thereof.
This percentage will not be specified in the present document, but
should be the subject of agreements between the network operators
and the program providers.
5.4.4a Severely Errored Second
(SES):
A specific case of the SETI where the given time interval is one
second.
5.4.5 Unavailable Time UAT A start of a period of Unavailable Time can be defined as
- either the onset of N consecutive SES/ SETI events; or
- the onset of a rolling window of length T in which M SES/
SETI events occur.
These time intervals/ seconds are considered to be part of the
Unavailable Time.
A end of period of Unavailable Time can be defined accordingly as
- the onset of N consecutive non-SES/ SETI events; or
- the onset of a rolling window of length T in which no SES/
SETI events occur.
These time intervals/ seconds are considered to be part of Available
Time.
The values N, M and T could differ for different types of service
(video, audio, data, etc.).
Note that these tests are only possible if Reed-Solomon encoding was used upstream with respect to the measurement
point.
5.5 Evaluation of service performance by combination of TS
related parameters
Introduction
Over the last years, numerous field trials were performed in the framework of research projects (see note) focused on
Quality of Service in digital TV. This applied to various types of digital TV networks such as satellite, cable, terrestrial,
and to a certain degree ATMnetworks. The trials aimed at creating artificially severe but realistic conditions for the
reception of the services. The supervision system created a database by collecting the measured parameters from the
measurement tools (RF parameters, TS analysis and audio and video perceived quality evaluator) located in different
points of the networks.
NOTE: ACTS Projects QUOVADIS (1995-1998) and MOSQUITO (1998-1999).
The statistical analysis of these data (representing the behaviour of the networks, measurement equipment and
supervision tools under realistic conditions) revealed certain correlations between individual parameters. A
methodology was defined by identifying a minimum set of parameters which describe in a consistent way the situation
for the receiving equipment in certain receiving conditions.
The definitions given hereafter are based on parameters which are already defined in this document, recommending a
suitable combination of such parameters to give a first approximation of the probability for a certain percentage of time
and location that a service is available in a certain area with a defined quality.
The aim is to provide the information in a structured form so that network operators can implement the functionalities
and gain experience with the measurement of the combined parameters. This could lead to a common understanding of
problems and potential solutions for the monitoring of Quality of Service, for example.
ETSI
35 ETSI TR 101 290 V1.2.1 (2001-05)
This could also be a potentially important feature for the definition of contractual obligations between service provider
and network operator. For a first estimate of the quality of service available under certain receiving conditions, the
parameters Service_Availability_Error, Service_Degradation_Error, and Service_Impairments_Error could be evaluated
and their level could be compared for a certain percentage of time with the predefined target value (as set, for example,
by the network operator).
5.5.1 Service_Availability_Error and Service_ Availability _Error_Ratio
Purpose To identify severe distortions and interruptions of the service under certain receiving
conditions. The parameter is related to the loss of the service.
Interface Z
Method Count the occurence of error messages for the following parameters over a defined time
interval ΔΤ (e. g. 10 s):
1) TS_sync_loss (see 5.2.1 {1.1})
2) PAT_error (see 5.2.1 {1.3})
3) PMT_error (see 5.2.1 {1.5})
For each time interval ΔΤ, the following differences are calculated (which correspond to the
derivation of the increasing function related to the occurrence of the concerned error
messages):
TS_sync_loss (ΔΤ) = TS_sync_loss (T) - TS_sync_loss (Τ−ΔΤ)
PAT_error (ΔΤ) = PAT_error (T) - PAT_error (Τ−ΔΤ)
PMT_error (ΔΤ) = PMT_error (T) - PMT_error (Τ−ΔΤ)
Then Service_Availability_Error value is calculated:
Service_Availability_Error = Max [TS_sync_loss (ΔΤ), PAT_error (ΔΤ), PMT_error (ΔΤ) ]
and display the results over an appropriate period, e. g. 10 minutes, and calculate
Service_Availability_Error_Ratio as the percentage of time for which the parameter exceeds a
pre-defined threshold.
5.5.2 Service_Degradation_Error and Service_Degradation_Error_Ratio
Purpose To identify severe degradation under certain receiving conditions. This parameter is related to
the level of strong impairments of the service.
Interface Z
Method Count the occurence of error messages for the following parameters over a defined time
interval ΔΤ (e. g. 10 s):
1) CRC_error (see 5.2.2 {2.2})
2) PCR_error (see 5.2.2 {2.3})
3) NIT_error (see 5.2.3 {3.1})
4) SDT_error (see 5.2.3 {3.5})
For each time interval ΔΤ, the following differences are calculated (which correspond to the
derivation of the increasing function related to the occurrence of the concerned error
messages):
CRC_error (ΔΤ) = CRC_error (T) - CRC_error
PCR_error (ΔΤ) = PCR_error (T) - PCR_error (Τ−ΔΤ)
NIT_error (ΔΤ) = NIT_error (T) - NIT_error (Τ−ΔΤ)
SDT_error (ΔΤ) = SDT_error (T) - SDT_error (Τ−ΔΤ)
Then Service_Degradation_Error value is calculated:
Service_Degradation_Error = Max [CRC_error (ΔΤ), PCR_error (ΔΤ), NIT_error (ΔΤ),
SDT_error (ΔΤ)]
and display the results over an appropriate period, e. g. 10 minutes, and calculate
Service_Degradation_Error_Ratio as the percentage of time for which the parameter exceeds
a pre-defined threshold.
ETSI
36 ETSI TR 101 290 V1.2.1 (2001-05)
5.5.3 Service_Impairments_Error and Service_Impairments_Error_Ratio
Purpose To identify first signs of service degradation under certain receiving conditions. The parameter
is related to unfrequent impairments of the service.
Interface Z
Method Count the occurence of error messages for the following parameter over a defined time
interval ΔΤ (e. g. 10 s):
1. Continuity_count_error (see 5.2.1 {1.4})
2. Transport_error (see 5.2.2 {2.1})
For each time interval ΔΤ, the following differences are calculated (which correspond to the
derivation of the increasing function related to the occurrence of the concerned error
messages):
Continuity_count_error (ΔΤ) = Continuity_count_error (T) - Continuity_count_error (Τ−ΔΤ)
Transport_error (ΔΤ) = Transport_error (T) - Transport_error (Τ−ΔΤ)
Then Service_Impairments_Error value is calculated:
Service_Impairments_Error = Max [Continuity_count_error, Transport_error]
and display the results over an appropriate period, e. g. 10 minutes, and calculate
Service_Impairments_Error_Ratio as the percentage of time for which the parameter exceeds
a pre-defined threshold.
An example for the definition of different reception conditions could be:
very good reception quality
(pQoS), no visible or audible
impairments for several minutes
Service_Availability_Error at Performance Class = 1 for 100 % of the time,
Service_Degradation_Error at Performance Class = 1 for 100 % of the time,
Service_Impairments_Error at Performance Class <= 2 for 95 % of the time
very bad reception conditions Service_Availability_Error at Performance Class >= 2 for 75 % of the time,
Service_Degradation_Error at Performance Class >= 2 for 95 % of the time,
Service_Impairments_Error at Performance Class >= 3 for 95 % of the time
NOTE: The figures in this example are not generally applicable. They may be defined by network operators or
service providers to quantify availability and/ or performance of a service in contractual agreements. In
addition, large variations of the figures are likely for different types of services.
For the purpose of these measurements, it may be useful to define several performance classes in relation with the
perceived Quality-of-Service (pQoS).
Hereafter an example is given that may be used for video and audio services:
Performance Class 1: high perceived Quality of Service (pQoS), no distortions.
Performance Class 2: good pQoS, few impairments.
Performance Class 3: low pQoS, repeated impairments.
Performance Class 4: very low pQoS, repeated interruptions of services.
Performance Class 5: repeated loss of service, impossible to follow any programme.
5.6 Parameters for CI related applications
The Common Interface (CI) is - in principle - a Transport Stream interface but it has particular properties which require
additional tests.
The parameters defined in this clause are intended to enable reproducible and comparable measurements on the CI. As
in the previous clauses on Transport Stream related tests and measurements, it cannot be assumed that these tests
provide a complete analysis. They are also designed as a 'health check', not as an overall compliance or conformance
test.
ETSI
37 ETSI TR 101 290 V1.2.1 (2001-05)
The following reference model pictures the interfaces and the functional blocks which are referred to in the definitions
of the tests.
TS(204)
inv. FEC
TS(188)
Gap
Inserter Module 1
TS
descrambled
to Demux
(part of Demux) and Decoder
Host internal external
Module 2
Module n
A1
B1
A2
B2
An
Bn
Co
Cn
C1
C2
GAP
Extractor
Cn-1
Typical CI Module
Bx
Ax
Gap
Processor
Bx
Ax Bypass
Input
Sidecar
Figure 5.5: CI Reference model
5.6.1 Latency
Parameter Purpose Interface Method
Latency To determine the impact of
one CI module on latency (or
average delay),
An - Bn Measure arrival time of synch
bytes of corresponding TS packets
at both interfaces;
ETSI
38 ETSI TR 101 290 V1.2.1 (2001-05)
5.6.2 CI_module_delay_variation
Parameter Purpose Interface Method Refernce
CI_module_delay_
variation
To check compliance with
CI spec,
to limit additional PCR jitter
and support decodability
Ax - Bx measure delay for all
corresponding bytes of each TS
packet between input Ax and
output Bx and calculate peak
delay variation for each TS
packet;
EN 50221 [23],
clause 5.4.2
NOTE: Ax and Bx are the input and output of any one CI Module.
5.6.3 Input_output_TS comparison
Parameter Purpose Interface Method
Input-output TS
comparison
To ensure that modules
under test do not impair other
parts of the TS
Co - Cn TS with at least 1 PID unaffected
by the CI modules + other PIDs
which will activate each module
under test and carry out a bitwise
comparison for the unaffected
PIDs;
additionally the CI modules should
be tested while inactive.
5.6.4 CI_module_throughput
Parameter Purpose Interface Method Limits
Period between
consecutive synch
bytes
To ensure compliance with
CI spec
Ax, Bx or
Cx
Measure time between 2 synch
bytes after processing in modules
@Ax: modules able to accept
input TS
@Bx: module outputs TS within
limits
58 Mbit/s from
EN 50221 [23]
NOTE: Ax and Bx are the input and output of any one CI Module, Cx is any corresponding interface of the host
device.
5.6.5 Valid TS on CI
Parameter Purpose Interface Method Limits
Valid TS To ensure decodability Ax, Bx or
Cx
Checks as in ETR 290 [21] 1st
priority + 2.6
NOTE: Ax and Bx are the input and output of any one CI Module, Cx is any corresponding interface of the host
device.
ETSI
39 ETSI TR 101 290 V1.2.1 (2001-05)
6 Common parameters for satellite and cable
transmission media
6.1 System availability
Purpose: The system availability describes the long-term quality of the complete digital transmission system
from MPEG-2 encoder to the measurement point.
Interface Z
Method: The definition of System Availability is based on the list of performance parameters of table 5.4:
Severely Disturbed Period (SDP)
Errored Block (EB)
Errored Time Interval ETI/ Errored
Second (ES)
Severely Errored Time Interval SETI/
Severely Errored Second (SES)
Unavailable Time UAT
The System Avalability is defined as the ratio of (Total Time - Unavailable Time) to Total Time.
6.2 Link availability
Purpose The link availability describes the long term quality of a specified link in a digital transmission chain. It
could be used as a quality of service parameter in contracts between network operators and program
providers.
Interface X (Overload indicator of the Reed Solomon decoder).
Method The definition of Link availability is based on following performance parameters:
Uncorrectable Packet (UP) An MPEG-2 TS packet with an uncorrectable error, which is
indicated by overload at the Reed-Solomon decoder.
Uncorrectable Time Interval UTI/
Uncorrectable Second (US)
A given time interval with one or more UPs.
The US is a specific case of theUTI where the given time interval
is one second.
Severely Uncorrectable Time Interval
(SUTI)/ Severely Uncorrectable
Second (SUS):
A given time interval which contains greater than a specified
percentage of Uncorrectable Packets, or at least one SDP (see 5.4)
or part thereof.
NOTE: This percentage will not be specified in the present
document, but should be the subject of agreements
between the network operators and the service
providers.
The SUS is a specific case of the SUTI where the given time
interval is one second.
ETSI
40 ETSI TR 101 290 V1.2.1 (2001-05)
Link Unavailable Time LUAT A start of a period of Link Unavailable Time can be defined as:
- either the onset of N consecutive SUS/ SUTI events; or
- the onset of a rolling window of length T in which M SUS/ SUTI
events occur.
These time intervals/ seconds are considered to be part of the Link
Unavailable Time.
A end of period of Link Unavailable Time can be defined
accordingly as:
- the onset of N consecutive non-SUS/ SUTI events; or
- the onset of a rolling windowof length T in which no SUS/ SUTI
events occur.
These time intervals/ seconds are considered to be part of Link
Available Time.
The values N, Mand T could differ for different types of service
(video, audio, data, etc.).
6.3 BER before RS decoder
Purpose The Bit Error Rate (BER) is the primary parameter which describes the quality of the digital
transmission link.
Interface W
Method The BER is defined as the ratio between erroneous bits and the total number of transmitted bits.
Two alternative methods are available; one for "Out of Service" and a second for "In Service" use. In
both cases, the measurement should only be done within the "link available time" as defined in
clause 6.2.
6.3.1 Out of service
The basic principle of this measurement is to generate within the channel encoder a known, fixed, repeating sequence of
bits, essentially of a pseudo random nature. In order to do this the data entering the sync-inversion/ randomization
function is a continuous repetition of one fixed TS packet. This sequence is defined as the null TS packet in
ISO/IEC 13818-1 [1] with all data bytes set to 0x00. i.e. the fixed packet is defined as the four byte sequence 0x47,
0x1F, 0xFF, 0x10, followed by 184 zero bytes (0 x 00). Ideally this would be available as an encoding system option
(see clause A.2).
6.3.2 In service
The basic assumption made in this measurement method is that the RS check bytes are computed for each link in the
transmission chain. Under normal operational circumstances, the RS decoder will correct all errors and produce an
error-free TS packet. If there are severe error-bursts, the RS decoding algorithm may be overloaded, and be unable to
correct the packet. In this case the transport_error_indicator bit shall be set, no other bits in the packet shall be changed,
and the 16 RS check bytes shall be recalculated accordingly before re-transmission on to another link. The BER
measured at any point in the transmission chain is then the BER for that particular link only.
The number of erroneous bits within a TS packet will be estimated by comparing the bit pattern of this TS packet before
and after RS decoding. If the measured value of BER exceeds 10-3 then the measurement should be regarded as
unreliable due to the limits of the RS decoding algorithm. Any TS packet that the RS decoder is unable to correct
should cause the calculation to be restarted.
6.4 Error events logging
Purpose Error events logging creates a permanent error log which can subsequently be used to locate possible
sources of errors. It may be used as a measure of "system availability" (see clause 6.1 above).
Interface Z
ETSI
41 ETSI TR 101 290 V1.2.1 (2001-05)
Method Loss of sync, loss of signal, and reception of errored TS packets are logged.
In case of sync or signal loss, the absolute time of loss shall be recorded, along with either the duration
of loss or the time of recovery from loss. A default time resolution of 1 second is strongly
recommended for this measurement, but other time intervals may be appropriate depending on the
application.
In case of reception of EBs (see clause 6.1), the number of such events in each second shall be logged,
together with the PID and the total number of received packets of this PID within the resolution time.
Logging of any other parameters (e.g. overloading of Reed-Solomon decoder, original_network_id,
service_id) are optional.
The error log shall store the most recent 1 000 error events as a minimum. Provision should be made to
access all of the error information in a form suitable for further data processing.
6.5 Transmitter symbol clock jitter and accuracy
Purpose Inaccuracies of the symbol clock concerning absolute frequency, frequency drift and jitter may
introduce intersymbol interference. Additionally, the accuracy of transmitted clock references like the
Program Clock Reference (PCR) can be influenced. Therefore the degradation of signal quality due to
symbol clock inaccuracies has to be negligible. Symbol clock jitter and accuracy can be degraded if the
symbol clock is directly synthesized from an unstable TS data clock. For this reason, the measurement
should be performed while the transmitter is driven by a TS to ensure a worst case measurement is
obtained.
Interface E
Method For measurements the absolute frequency, frequency wander and timing jitter are of interest. A PLL
circuit can be used for synchronization to the symbol clock and according to the loop bandwidth,
timing jitter is suppressed and low frequency drift (wander) is still present at the output of the loop
oscillator. Jitter can be measured with an oscilloscope by triggering with the extracted clock. Jitter is
usually expressed as a peak-to-peak value in UI (Unit Interval) where one UI is equal to one clock
cycle (Tsymbol). For measurements of the absolute frequency and frequency wander the output of the
clock extractor can be used or the symbol clock directly using an appropriate frequency counter.
NOTE: This measurement refers to the physical layer of TS interconnection. See clause 5.3.2 for
PCR measurements.
6.6 RF/IF signal power
Purpose Level measurement is needed to set up a network.
Interface Any RF/IF interface, N, P.
Method The signal power, or wanted power, is defined as the mean power of the selected signal as would be
measured with a thermal power sensor. Care should be taken to limit the measurement to the bandwidth
of the wanted signal. When using a spectrum analyser or a calibrated receiver, it should integrate the
signal power within the nominal bandwidth of the signal (symbol rate x(1 + α)).
6.7 Noise power
Purpose Noise is a significant impairment in a transmission network.
Interface N (out of service) or T (in service)
Method The noise power (mean power), or unwanted power, is measured with a spectrum analyser (out of
service) or an estimate is obtained from the IQ diagram (in service), see clause 6.9.9. The noise level is
specified using either the occupied bandwidth of the signal, which is equal to the symbol rate x (1 + α).
ETSI
42 ETSI TR 101 290 V1.2.1 (2001-05)
See annex G.
6.8 Bit error count after RS
Purpose To measure whether the MPEG-2 TS is quasi error free.
Interface Z
Method The same principle as used for the "Out of service measurement" of the "BER before the
Reed-Solomon decoder" described in clause 6.3.2, with the modification that the result is presented as
an error count rather than a ratio. The receiver only has to compare the received TS packets with the
Null packets as defined in clause A.2.
6.9 IQ signal analysis
6.9.1 Introduction
Assuming:
- a constellation diagram of M symbol points; and
- a measurement sample of N data points, where N is sufficiently larger than M to deliver the wanted measurement
accuracy; and
- the co-ordinates of each received data point j being Ij + δIj, Qj + δQj where I and Q are the co-ordinates of the
ideal symbol point and δI and δQ are the offsets forming the error vector of the data point (see clause A.3).
Figure 6-1: Relationship between the parameters describing different IQ distortions
Modulation Error Ratio (MER) and the related Error Vector Magnitude (EVM) are calculated from all N data points
without special pre-calculation for the data belonging to theM symbol points.
With the aim of separating individual influences from the received data, for each point i of the M symbol points the
mean distance di and the distribution σi can be calculated from those δIj, δQj belonging to the point i.
From the M values {d1, d2, ... dM} the influences/parameters:
- originoffset;
- amplitude Imbalance (AI); and
- quadrature Error (QE),
ETSI
43 ETSI TR 101 290 V1.2.1 (2001-05)
can be extracted and removed from the di values, allowing to calculate the Residual Target Error (RTE) with the same
algorithm as the System Target Error (STE) from {d1, d2, ... dM}.
From the statistical distribution of the M clouds (denoted by σi in figure 6-2) parameters:
- phase jitter; and
- CWinterferer,
may be extracted. The remaining clouds (after elimination of the above two influences) are assumed to be due to
Gaussian noise only and are the basis for calculation of the signal-to-noise ratio. The parameter may include - besides
noise - also some other disturbing effects, like small non-coherent interferers or residual errors from the equalizer. From
the SNR value the Carrier/Noise value can be estimated (see clause A.3).
When using the interfaces E or G filtering of the signal before the interface should be considered.
6.9.2 Modulation Error Ratio (MER)
Purpose To provide a single "figure of merit" analysis of the received signal.
This figure is computed to include the total signal degradation likely to be present at the input of a
commercial receiver's decision circuits and so give an indication of the ability of that receiver to
correctly decode the signal.
Interface E, G, S, T
Method The carrier frequency and symbol timing are recovered, which removes frequency error and phase
rotation. Origin offset (e.g. cause by residual carrier or DC offset), quadrature error and amplitude
imbalance are not corrected.
A time record of N received symbol co-ordinate pairs ( ) I j Q j
~
,
~ is captured.
For each received symbol, a decision is made as to which symbol was transmitted. The ideal position of
the chosen symbol (the centre of the decision box) is represented by the vector ( ) I j ,Q j . The error
vector ( ) δ I j ,δQ j is defined as the distance from this ideal position to the actual position of the
received symbol.
In other words, the received vector ( ) I j Q j
~
,
~
is the sum of the ideal vector ( ) I j ,Q j and the error
vector ( ) δ I j ,δQ j .
The sum of the squares of the magnitudes of the ideal symbol vectors is divided by the sum of the
squares of the magnitudes of the symbol error vectors. The result, expressed as a power ratio in dB, is
defined as the Modulation Error Ratio (MER).
( )
( )
dB
I Q
I Q
MER
N
j
j j
N
j
j j










+
+
= ×


=
=
1
2 2
1
2 2
10 log10
δ δ
The definition of MER does not assume the use of an equalizer, however the measuring receiver may
include a commercial quality equalizer to give more representative results when the signal at the
measurement point has linear impairments.
When an MER figure is quoted it should be stated whether an equalizer has been used.
It should be reconsider that MER is just one way of computing a "figure of merit" for a vector
modulated signal. Another "figure of merit" calculation is Error Vector Magnitude (EVM) defined in
clause A.3. It is also shown in clause A.3 that MER and EVM are closely related and that one can
generally be computed from the other.
ETSI
44 ETSI TR 101 290 V1.2.1 (2001-05)
MER is the preferred first choice for various reasons itemized in clause A.3.
6.9.3 System Target Error (STE)
Purpose The displacement of the centres of the clouds in a constellation diagram from their ideal symbol point
reduces the noise immunity of the system and indicates the presence of special kind of distortions like
Carrier Suppression, Amplitude Imbalance, Quadrature Error (QE) and e.g. non-linear distortions. STE
gives a global indication about the overall distortion present on the raw data received by the system.
Interface E, G, S, T
Method For each of theM symbol points in a constellation diagram compute the distance di between the
theoretical symbol point and the point corresponding to the mean of the cloud of this particular symbol
point. This quantity ( di ) is called Target Error Vector (TEV) and is shown in figure 6-2.
Figure 6-2: Definition of Target Error Vector (TEV)
From the magnitude of the MTarget Error Vectors calculate the mean value and the standard deviation
(normalized to Srms , defined as the RMS amplitude value of the points in the constellation), obtaining
the System Target Error Mean (STEM) and the System Target Error Deviation (STED) as follows:
( )
=
= +
N
j
rms I j Qj
N
S
1
1 2 2
=
×
=
M
i
i
rms
d
M S
STEM
1
1
2
2
1
2
STEM
M S
d
STED
rms
M
i
i
−
×
=
=
ETSI
45 ETSI TR 101 290 V1.2.1 (2001-05)
6.9.4 Carrier suppression
Purpose A residual carrier is an unwanted coherent CW signal added to the QAM signal. It may have been
produced by DC offset voltages of the modulating I and/or Q signal or by crosstalk from the
modulating carrier within the modulator.
Interface E, G, S, T
Method Search for systematic deviations of all constellation points and isolate the residual carrier. Calculate the
Carrier Suppression (CS) from the formula:






= ×
RC
sig
P
P
CS 10 log10
where PRC is the power of the residual carrier and Psig is the power of the QAM signal (without
residual carrier).
6.9.5 Amplitude Imbalance (AI)
Purpose To separate the QAMdistortions resulting from AI of the I and Q signal from all other kind of
distortions.
Interface E, G, S, T
Method Calculate the I and Q gain values vI and vQ from all points in a constellation diagram eliminating all
other influences. Calculate AI from νI and νQ:
min( , ) and max( , ) .
1 100 %
1 2
1
2
with v vI vQ v vI vQ
v
v
AI
= =
×





= −
( )
( )
( )
( )
( ) ( ) i I i Q i
N
j
i Q j
M
i i
i i Q
Q
N
j
i I j
M
i i
i i I
I
d d d
Q
N
d
Q
Q d
M
I
N
d
I
I d
M
+ =
=
+
=
=
+
=




=
=
=
=
(Q- component of di as given in subclause 6.9.3)
(I - component of di as given in subclause 6.9.3)
1
1
1
1
1
1
1
1
δ
ν
δ
ν
6.9.6 Quadrature Error (QE)
Purpose The phases of the two carriers feeding the I and Q modulators have to be orthogonal. If their phase
difference is not 90° a typical distortion of the constellation diagram results. The receiver usually aligns
its reference phase in such a way that the 90° error (Δϕ) is equally spread between ϕ1 and ϕ2.
ETSI
46 ETSI TR 101 290 V1.2.1 (2001-05)
I
Q
Decision Boundary
Signal Point
Decision Boundary Box
1 ϕ
2 ϕ
90°-QE
Figure 6-3: Distortion of constellation diagram resulting from I/Q Quadrature Error (QE)
Interface E, G, S, T
Method Search for the constellation diagram error shown in figure 6-3 and calculate the absolute value of the
phase difference Δϕ = |ϕ1 - ϕ2| after having eliminated all other influences and convert this into
degrees.
= ° × 1 − 2 [°]
180 ϕ ϕ
π
QE
6.9.7 Residual Target Error (RTE)
Purpose The RTE is a subset of the distortions measured as System Target Error (STE) with influences of
Carrier Suppression, Amplitude Imbalance, and Quadrature Error (QE) removed. The remaining
distortions may result mainly from non-linear distortions.
Interface E, G, S, T
Method Remove from the Target Error Vectors di, which have been used to calculate the Symbol Target Error
(STE), the influences of Carrier Suppression, Amplitude Imbalance, and Quadrature Error (QE), call
the remaining vectors d'i and calculate the mean value of their magnitudes.
=
′
×
=
M
i
i
rms
d
M S
RTE
1
1
6.9.8 Coherent interferer
Purpose Coherent interferers (not necessarily related to the main carrier) are usually measured with a spectrum
analyser (out of service, and in some cases in service with narrow resolution bandwidth filter and video
filter at interfaces N and P) or either of the following methods described below (in service). In a
constellation diagram a sine-wave interferer will change the noisy clouds of each system point into a
"donut" shape. From the statistical distribution of the clouds, the amplitude of the interferer can be
calculated if it is above a certain limit. If the frequency of the interferer is of interest or more than one
interferer is present, the Fourier transform method should to be used.
Interface E, G, S, T
Method Perform a Fourier transform of a time record of error vectors to produce a frequency spectrum of the
interferers.
ETSI
47 ETSI TR 101 290 V1.2.1 (2001-05)
Alternatively, calculate the RMS magnitude ai of the coherent interferer preferably from the statistical
distribution of the 4 inner clouds computed from the measurement sample. Normalize ai to Srms and
express the result in dB.
i
rms
a
S
C / I = 20× log10 [dB]
NOTE 1: In the present document, the term "coherent" is applied to signals that have a high degree of correlation
with a time shifted version of itself.
EXAMPLE 1: Continuous Waves (CW) or even single channel analogue video modulated carriers, these signals
are coherent although they do not need to be related to the carrier of the digital channel under test.
NOTE 2: Non-coherent is applied to signals with very low correlation to a time shifted version of themselves.
EXAMPLE 2: Random noise or digitally modulated carriers, as well as the combined result of inter-modulation
by many carriers.
6.9.9 Phase Jitter (PJ)
Purpose The PJ of an oscillator is due to fluctuations of its phase or frequency. Using such an oscillator to
modulate a digital signal results in a sampling uncertainty in the receiver, because the carrier
regeneration cannot follow the phase fluctuations.
The signal points are arranged along a curved line crossing the centre of each decision boundary box as
shown in figure 6-4 the four "corner decision boundary boxes".
Q
I
"Corner Decision Boundary Box"
for calculation of the Phase Jitter
Arc section through a
Figure 6-4: Position of arc section in the constellation diagram to define the PJ
(example: 64-QAM)
Interface E, G, S, T
Method Phase Jitter (PJ) can be calculated theoretically using the following algorithm:
For every received symbol:
1) Calculate the angle between the I-axis of the constellation and the vector to the received symbol
)
~
,
~
(I Q :
I
Q~
~
φ1 = arctan
2) Calculate the angle between the I-axis of the constellation and the vector to the corresponding ideal
ETSI
48 ETSI TR 101 290 V1.2.1 (2001-05)
symbol (I ,Q) :
I
Q
φ 2 = arctan
3) Calculate the error angle:
φ E =φ1 −φ2
From these N error angles calculate the RMS phase jitter:

= =






= −
N
i
N
i
N Ei N Ei
PJ
1
2
1
2
1 2 1 φ φ
However, the followingmethodmay be more practical.
The first approximation of the "arc section" of a "corner decision boundary box" is a straight line
parallel to the diagonal of the "decision boundary box". Additionally the curvature of the Phase Jitter
(PJ) trace has to be taken into account when calculating the standard deviation of the PJ. The mean
value of the PJ is calculated in degrees.
( )






× − ×
= ° ×
M d
PJ PJ
2 1
arcsin
180 σ
π
where M = order of QAM and 2d = distance between two successive boundary lines.
Within the argument of the arc sine function, the standard deviation of the PJ is referenced to the
distance from the centre of the "corner decision boundary box" to the centre point of the QAM signal.
6.9.10 Signal-to-Noise Ratio (SNR)
Purpose see 6.9.1
Interface S, T
Method see 6.9.1, G.8, A.3
6.10 Interference
Purpose In a CATV network interference products can be caused by modulators and frequency converters.
Interface N (out of service) or S, T (in service).
Method Out of service interference products are measured with a spectrum analyser and in some cases inservice
measurements can be done if a narrow resolution bandwidth filter and video filtering is used to
lower the response of the instrument to the signal spectrum. If the frequency of the expected
interference is known, the measurement can be made easily and quickly. In-service information of
coherent interference can be derived from the constellation, clause 6.9.8.
In some circumstances the residual carrier level can be measured with a spectrum analyser, by using a
narrow resolution bandwidth filter and video filtering, at the interfaces H, J, N, P. The CS can be
calculated as ten times the logarithm (base 10) of the ratio of the signal power measured as described in
clause 6.6, to the measured remaining carrier power.
ETSI
49 ETSI TR 101 290 V1.2.1 (2001-05)
7 Cable specific measurements
In SMATV networks that distribute the 1st satellite IF directly to subscribers, some parameters of this clause can be
defined accordingly for QPSK modulated signals.
7.1 Noise margin
Purpose To provide an indication of the reliability of the transmission channel. The noise margin measurement
is a more useful measure of system operating margin than a direct BER measurement due to the
steepness of the BER curve.
Interface The reference interface for the noise injection is the RF interface (N). For practical implementation,
other interfaces can be used, provided equivalence can be shown, for example P.
Method The noise margin is computed by adding white Gaussian noise on the received signal. The noise
margin will be the difference in dB between the carrier to noise ratio (C/N) of the received signal and
the carrier to noise ratio for a BER of 10-4 (before RS decoding).
7.2 Estimated noise margin
Purpose To provide an indication of the reliability of the transmission channel without switching off the service.
The noise margin measurement is a more useful measure of system operating margin than a direct BER
measurement due to the steepness of the BER curve.
Interface T
Method The estimated noise margin is computed by simulating the addition of white Gaussian noise to the
demodulated data and predicting the resulting BER by statistical methods.
The noise margin will be the difference in dB between the estimated SNR of the received signal and
the synthesized SNR which gives a predicted BER of 10-4 (before RS decoding).
7.3 Signal quality margin test
Purpose A fast and simple pass/fail measurement that can provide an indication of the quality of the digital
service at various nodes in the cable distribution network.
This measurement will provide a first indication of the margin to failure of the digital service. It can be
used as a signal quality check during installation, and as a maintenance tool for basic monitoring of
signal quality through the network.
Interface T. The measurement assumes the use of an equalizer.
Method The demodulated, equalized and sampled IQ constellation characteristically has data points clustered
around each of the ideal data point locations. For a high quality signal, most of the received data points
are close to the ideal location and the clusters' spread is small relative to the overall constellation size.
As the signal is degraded by noise and other impairments the clusters' spreading increases leading to a
corresponding increase in symbol errors as more data points stray over the inter-symbol decision
boundaries. In general, the amount of spread in the received data points is an indication of the signal
quality.
To measure the amount of data point spreading in the received constellation we place decision
boundaries to the left, right, above and below each constellation point. These boundaries form a
"quality threshold" box around each constellation point. The edges of this box are closer to the ideal
data point than the inter-symbol decision boundaries so a significant proportion of the received data
points may lie outside the quality threshold box even under normal conditions.
At all constellation points, the number of data points falling inside and outside the quality threshold
box are counted in order to compute a percentage which is then used to trigger the pass/fail indication.
ETSI
50 ETSI TR 101 290 V1.2.1 (2001-05)
Since the acceptable spread will vary depending on the point of measurement within the network, the
size of the quality threshold box is user selectable from a small range of sizes. For example, a small
quality threshold box for measurements at the head-end, a larger quality threshold box for
measurements at the customers premises.
The individual quality threshold box sizes are chosen by the network operator to give the same pass/fail
threshold at each measurement point in the network taking into account the signal degradation expected
under normal operating conditions.
The choice of threshold percentage and likely quality threshold box, the relationship between signal
quality margin and the critical BER of 10-4, the definition of an appropriate equalizer (see clause A.3),
and the possibility to include linear distortions in this measurement are all subject to further study.
Figure 7-1: Quality thresholds for single constellation in the I/Q plane
A single constellation point in the I/Q plane is shown in figure 7-1. Different quality thresholds can be
defined within the normal decision boundaries.
7.4 Equivalent Noise Degradation (END)
Purpose END is a measure of the implementation loss caused by the network or the equipment where the reference is
the ideal performance.
Interface T (BER) and N or P or R (noise injection)
Method The END is obtained from the difference in dB of the C/N or Eb/N0 ratio needed to reach a BER of 10-4 and
the C/N or Eb/N0 ratio that would theoretically give a BER of 10-4, for a Gaussian channel.
ETSI
51 ETSI TR 101 290 V1.2.1 (2001-05)
Figure 7-2: Measurement of equivalent noise degradation
Figure 7-2 is not the true theoretical curve representing BER in DVB-C systems, but only an example. This
figure will be updated by true theoretical values and, if necessary, tables corresponding to these values will
be given in an annex to the present document, when available. The theoretical curve in this figure needs to be
ETSI
52 ETSI TR 101 290 V1.2.1 (2001-05)
updated from data in the table contained in annex D.
7.5 BER vs. Eb/N0
Purpose The BER vs. Eb/N0 measurement enables a graph to be drawn which shows the implementation loss of
the system over a range of Bit Error Rates. The residual BER at high Eb/N0 values is an indicator of
possible network problems. C/N measurements can be converted to Eb/N0 as shown
f m
BW
N
C
N
E
s
b noise
×
= + 10
0
10log [in dB]
m is the number of bits per symbol (m = 6 for 64-QAM) and N is measured in the Nyquist bandwidth
(symbol rate as indicated in clause 6.7).
Interface T (BER) and N or P or R (noise injection)
Method The BER vs. Eb/N0 curve will be measured using the RF and noise power measurements described
above. The BER range of interest is 10-7 to 10-3. The Eb/N0 value is based on the gross bitrate
(including RS error correction) and the net bitrate value of Eb/N0 can easily be calculated using the RS
rate, using the following conversion factor for a RS (204, 188) code (see annex G).
0,35 dB
188
204
log 10 10 + =




×
7.6 Phase noise of RF carrier
Purpose Phase noise can be introduced at the transmitter side or by the receiver due to unstable local oscillators.
Phase noise outside the loop bandwidth of the carrier recovery circuit leads to a circular smearing of
the constellation points in the I/Q plane. This reduces the operating margin (noise margin) of the
system and may directly increase the BER.
Interface Any RF/IF interface, N, P
Method Phase noise power density is normally expressed in dBc/Hz at a certain frequency offset from the
carrier. Out of service phase noise will be measured with a spectrum- or modulation- analyser.
7.7 Amplitude, phase and impulse response of the channel
Purpose Linear distortions, like amplitude and phase response errors and echoes, will be caused for instance by
long lengths of cable and the cascading of a high number of amplifiers. The impulse response is
important to localize the discrete reflections that may occur in cable networks.
Interface S, T
Method The impulse response of the transmission channel can be calculated (inverse Fourier transform) from
the amplitude and phase response. The amplitude and phase response are defined as the RF-channel
response. The amplitude response of the transmission channel can be derived from the equalizer tap
coefficients or can be calculated directly from the "I" and "Q" samples, for example by using auto- and
cross-correlation functions.
ETSI
53 ETSI TR 101 290 V1.2.1 (2001-05)
7.8 Out of band emissions
Purpose To prevent interference in other channels in the network the RF signal shall comply with the spectrum
mask specified for the network under test.
Interface Transmitter output, J
Method Spectrum analyser
8 Satellite specific measurements
8.1 BER before Viterbi decoding
Purpose This measurement gives an indication of the transmission link performance. Due to typical error rates
ranging from 7 × 10-2 to 10-5 the measurement can be done in a reasonable amount of time. Outside of
this range the accuracy of the results may not be guaranteed.
Interface The measurement shall be done before the Viterbi decoder (Interface T of the receiver).
Method The signal after Viterbi decoding in the measurement instrument is coded again using the same coding
scheme as in the transmitter, in order to produce an estimate of the originally coded I and Q sequences.
These sequences are compared at bit level with the sign-values of the signals that are available before
Viterbi decoding.
The BER for the I and Q paths should be made available separately. The measurement should be based
on at least several hundred bit errors. For fast evaluation, in the case that the BER is lower than 10-4, it
should be possible to stop the measurement after approximately 1 second.
For accurate measurement of Eb/N0 at the quasi error free threshold, the measurement time and the
presentation of the result should be such that an accuracy of three decimal place can be achieved. The
quasi error free threshold corresponds to a BER before Viterbi decoding in the range 7 × 10-2 to
7 × 10-3,depending on the selected convolutional code rate; or a BER after Viterbi decoding of
2 × 10-4.
Figure 8-1: BER measurement before Viterbi decoding
8.2 Receive BER vs. Eb/No
Purpose To verify overall clear sky link performance and link margin using a reference down link for
acceptance tests.
Interface After Viterbi decoding, V
Method This is an out-of-service-measurement. The BER measurement shall be based on the null packets
inserted at the modulator as defined in clause A.1.
ETSI
54 ETSI TR 101 290 V1.2.1 (2001-05)
inserted at the modulator as defined in clause A.1.
To obtain the various values necessary for the curve BER over Eb/No, white Gaussian noise is injected
at the receiver site. In order to get accurate results it shall be verified that the inserted noise level is at
least 15 dB above the system noise. This can easily be observed on a spectrum analyser by switching
the inserted noise on and off. Stable reception conditions are a precondition for accurate measurement
results.
The RS decoding should be deactivated, or bypassed to avoid excessively long measurement periods.
The BERrange of interest is 10-9 to 10-2.
The measurement values are compared with the theoretical values. The value for the Equivalent Noise
Degradation (END) at a BER of 10-4 can be derived from this information as well.
For evaluation of Eb/No only the number of information bits (the net bitrate) shall be taken into
account.
8.3 IF spectrum
Purpose To prevent interference into other channels and to be compliant with the DVB specification the
modulator output spectrum shall be according with the one specified in EN 300 421 [5].
Interface H, input of the up-converter, typically 70 MHz or 140 MHz (Modulator output plus equipment for the
connection to the up-converter input).
Method Spectrum analyser and template for amplitude response, network analyser and template for group delay
response, both as specified in EN 300 421 [5].
9 Measurements specific for a terrestrial (DVB-T)
system
The intention of these guidelines is to provide a list of measurements useful in a DVB-T OFDM environment. The
different options could be selected by the users of the system. Equipment manufacturers (both transmitters and
receivers) as well as the operators, can choose those measurements that best fits their needs. A list of the applicability of
the measurement parameters described in the present document to the DVB-T transmitter, receiver and network is given
in the following table.
The measurements 6.1 "System availability" and 6.2 "Link availability" are also valid for Terrestrial (not only for Cable
and Satellite) and for any contribution link like SDH, PDH, etc.
ETSI
55 ETSI TR 101 290 V1.2.1 (2001-05)
Table 9-1: DVB-T measurement parameters and their applicability
Measurement parameter Transmitter Network Receiver
1) RF frequency measurements
1.1) RF frequency accuracy (Precision) X
1.2) RF channel width (Sampling Frequency Accuracy) X
1.3) Symbol Length measurement at RF (Guard Interval verification) X
2) Selectivity X
3) AFC capture range X
4) Phase noise of local oscillators (LO) X X
5) RF/IF signal power X X X
6) Noise power X
7) RF and IF spectrum X
8) Receiver sensitivity/ dynamic range for a Gaussian channel X
9) Equivalent Noise Degradation (END) X X
9a) Equivalent Noise Floor (ENF) X
10) Linearity characterization (shoulder attenuation) X
11) Power efficiency X
12) Coherent interferer X X
13) BER vs. C/N ratio by variation of transmitter power X X
14) BER vs. C/N ratio by variation of Gaussian noise power X X
15) BER before Viterbi (inner) decoder X X X
16) BER before RS (outer) decoder X X X
17) BER after RS (outer) decoder X X
18) I/Q analysis
18.1) N/A
18.2) Modulation Error Ratio X X X
18.3) System Target Error X X
18.4) Carrier Suppression X X
18.5) Amplitude Imbalance X X
18.6) Quadrature Error X X
18.7) Phase Jitter X X
19) Overall signal delay X X
20) SFN synchronization
20.1) MIP_timing_error X
20.2) MIP_structure_error X
20.3) MIP_presence_error X
20.4) MIP_pointer_error X
20.5) MIP_periodicity_error X
20.6) MIP_ts_rate_error X
21) System Error Performance X X X
ETSI
56 ETSI TR 101 290 V1.2.1 (2001-05)
Figure 9-1: Block diagram of a DVB-T transmitter
Figure 9-2: Block diagram of a DVB-T receiver
9.1 RF frequency measurements
The accuracy of some basic parameters of the OFDM modulation may be carried out at the RF layer of the DVB-T
signal.
9.1.1 RF frequency accuracy (Precision)
Purpose Successful processing of OFDM signals requires that certain carrier frequency accuracy be maintained
at the transmitter. Specific network operations modes such as SFN require high accuracy of the carrier
frequency.
Interface L, M
ETSI
57 ETSI TR 101 290 V1.2.1 (2001-05)
Method The 8k mode of the DVB-T always has a continual pilot, with continuous phase along successive
OFDM symbols, exactly at the channel centre (k = 3 408). Its frequency may be directly measured by
any spectrum analyser that has an integrated counter and at least a resolution filter of 300 Hz or less (if
necessary by utilizing a reference source of sufficient accuracy).
The 2k mode has a continual pilot with continuous phase at k = 1 140. Its frequency may be directly
measured by any spectrum analyser that has an integrated counter and at least a resolution filter of
300 Hz or less (if necessary by utilizing a reference source of sufficient accuracy). The centre channel
frequency may be inferred by subtracting to the measured frequency:
8 MHz channels: 1 285 714 Hz i.e. [(1 140 – 852) × 4 464,2 857 = 1 285 714 Hz].
7 MHz channels: 1 125 000 Hz i.e. [(1 140 – 852) × 3 906,25 = 1 125 000 Hz].
6 MHz channels: 964 286 Hz i.e. [(1 140 – 852) × 4464,2857 = 964 286 Hz].
NOTE: For 2k mode this method may have some inaccuracy if the sampling frequency of the
modulator is not precise, however such error in the sampling frequency would need to be
very high to significantly affect the centre channel measurement. Should more accuracy
needed, the two outer continual pilots may be measured as indicated under 9.1.2 RF
channel width, and the mean of the two values be calculated.
9.1.2 RF channel width (Sampling Frequency Accuracy)
Purpose Channel width measurements are convenient for verification that sampling frequency accuracy is
maintained at the modulator side.
Interface L, M
Method The occupied bandwidth of a COFDMmodulated channel depends directly from the frequency spacing
and this from the sampling frequency.
The outermost carriers in a DVB-T signal are continual pilot carriers. Their frequencies are measured
(see annex E.1) and the difference between them should be compared to the nominal channel width of
7 607 142,857 Hz for 8 MHz channels, 6 656 250,000 Hz for 7 MHz channels and 5 705 357,143 Hz
for 6 MHz channels.
NOTE: Three decimal places are given here for completeness only. Accuracy of 1 Hz at 5 MHz
means 0,2 × 10-6 per Hz, which may be enough for most cases of sampling frequency
measurement. Measurement instruments should have better accuracy and resolution
(typically in the order of ten times) than the required measurement accuracy.
If the frequency of the outermost carriers is known, see clauses E.1.3 and E.1.4 for how to measure
them, then the related values may be calculated as per table below. Denoting the outermost pilot
frequencies as FL and FH appropriately the occupied bandwidth is OB = FH _ FL. The number of
carriers is K, and for the 2k mode K-1 = 1 704 while for the 8k mode K-1 = 6 816.
Table 9.2: Calculated values
8k mode 2k mode
Occupied bandwidth FH - FL
Frequency Spacing (FH - FL)/6 816 (FH - FL)/1 704
Useful duration 6 816/(FH - FL) 1 704/(FH - FL)
Centre channel 1st IF (FH - FL) × 4 096/(K-1) (FH - FL) × 1 024/(K-1)
Sampling Frequency (FH - FL) × 16 384/(K-1) (FH - FL) × 4 096/(K-1)
ETSI
58 ETSI TR 101 290 V1.2.1 (2001-05)
9.1.3 Symbol Length measurement at RF (Guard Interval verification)
Purpose Verification of the guard interval used in a received DVB-T signal may be carried out at RF level by
carefull frequency measurements. This measurement is valid in cases where there is an uncertainty on
whether a modulator is correctly working and producing a signal with the expected or assigned Guard
Interval.
Interface L, M
Method The scattered pilots produce a pulsed-like spectrum every third carrier in a DVB-T spectrum due to
their repetition presence at the same phase and location every fourth symbol. The frequency difference
between two contiguous spectral lines representing a scattered pilot represents the inverse of the time
length of four consecutive DVB-T symbols.
Measuring such frequency difference and dividing its inverse by 4 will provide the total symbol length
TS of the measured signal. By subtracting the nominal useful symbol duration TU the length of the GI
is found. See annex E.1 for details on the measurement procedure and symbol lengths.
9.2 Selectivity
Purpose To identify the capability of the receiver to reject out-of-channel interference.
Interface The measurement of the signal input level and the interferer shall be carried out at the interface N,
using interface W or X for the BER monitoring.
Method The input power is adjusted to 10 dB above the minimum input power as defined in "Receiver
sensitivity" (see clause 9.8). The C/I threshold needed for QEF operation after RS decoder
(BER < 2 x 10-4 before RS decoder) should be measured as a function of the frequency of a CW
interferer.
9.3 AFC capture range
Purpose To determine the frequency range over which the receiver will acquire overall lock.
Interface N, for the application of the test signal; Z, for the test of TS synchronization
Method A signal is applied to the input of the receiver, at a level 10 dB above the minimum input power as
defined in "Receiver sensitivity" (see clause 9.8). The signal is frequency shifted in steps (from below
and above) towards a nominal value and the Sync_byte_error is verified according to clause 5.2.1
(Measurement and analysis of the MPEG-2 TS - First priority: necessary for decodability (basic
monitoring)).
9.4 Phase noise of Local Oscillators (LO)
Purpose Phase noise can be introduced at the transmitter, at any frequency converter or by the receiver due to
random pertubation of the phase of the oscillators.
In an OFDM system the phase noise can cause Common Phase Error (CPE) which affects all carriers
simultaneously, and which can be minimized or corrected by using the continual pilots. However the
Inter-Carrier Interference (ICI) is noise-like, cannot be corrected.
The effects of CPE are similar to any single carrier system and the phase noise, outside the loop
bandwidth of the carrier recovery circuit, leads to a circular smearing of the constellation points in the
I/Q plane. This reduces the operating margin (noise margin) of the system and may directly increase
the BER.
The effects of ICI are peculiar to OFDM and cannot be corrected for. This has to be taken into account
as part of the total noise of the system.
Interface Any access to Local Oscillators (LO), in transmitters, converters and receivers.
ETSI
59 ETSI TR 101 290 V1.2.1 (2001-05)
Method Phase noise can be measured with a spectrum analyser, a vector analyser or a phase noise test set.
Method for
CPE:
Phase noise power density is normally expressed in dBc/Hz at a certain frequency offset from the local
oscillator signal. It is recommended to specify a spectrum mask with at least three points (frequency
offsets and levels), for example see figure 9-3.
NOTE: See clauses A.4 and E.4 for additional information on phase noise measurements. See
clause E.4.1 for some practical information.
Carrier
Figure 9-3: Possible mask for CPE measurements
Method for
ICI:
For the measurement of ICI, the use of multiples of the carrier spacing is recommended for the
frequencies, fa, fb, fc.
Table 9.3: Frequency offsets for 2 k and 8 k systems
2 k system 4,5 kHz 8,9 kHz 13,4 kHz
8 k system 1,1 kHz 2,2 kHz 3,4 kHz
Typical use For manufacturing, incoming inspection and maintenance of modulators, transmitters, up/ down
converters and receivers, either professional or consumer type.
9.5 RF/IF signal power
Purpose Signal power, or wanted power, measurement is required to set and check signal levels at the
transmitter and receiver sites.
Interface K, L, M, N, P
Method The signal power of a terrestrial DVB signal, or wanted power, is defined as the mean power of the
signal as would be measured with a thermal power sensor. In the case of received signals care should
be taken to limit the measurement to the bandwidth at the wanted signal. When using a spectrum
analyser or a calibrated receiver, it should integrate the signal power within the nominal bandwidth of
the signal (n × fSPACING) where n is the number of carriers.
ETSI
60 ETSI TR 101 290 V1.2.1 (2001-05)
9.6 Noise power
Purpose Noise is a significant impairment in a transmission network.
Interface N,P
Method The noise power (mean power), or unwanted power, can be measured with a spectrum analyser (out of
service). The noise power is specified using the occupied bandwidth of the OFDM signal (n × fSPACING)
where n is the number of carriers.
NOTE: The term C/N should be calculated as the ratio of the signal power, measured as described in clause 9.5,
to the noise power, measured as described in this clause.
9.7 RF and IF spectrum
Purpose To avoid interfering with other channels, the transmitted RF spectrum should comply to a spectrum
mask, which is defined for the terrestrial network. If the spectrum at the modulator output is defined by
a spectrum mask, the same procedure can be applied to the IF signal (with no pre-correction active).
Interface K, M
Method This measurement is usually carried out using a spectrum analyser. The spectral density of a terrestrial
DVB signal is defined as the long-term average of the time-varying signal power per unity bandwidth
(i.e. 1 Hz). Values for other bandwidths can be achieved by proportional increase of the values for
unity bandwidth.
To avoid regular structures in the modulated signal a non-regular, e.g. a Pseudo-Random Binary
Sequence (PRBS) -like or a programme type digital transmitter input signal is necessary.
Care has to be taken that the input stage of the selective measurement equipment is not overloaded by
the main lobe of the signal while assessing the spectral density of the side lobes, i. e. the out-of-band
range. Especially in cases with very strong attenuation of the side lobes non-linear distortion in the
measurement equipment can produce side lobe signals that mask the original ones. Selective
attenuation of the main lobe has proven to be in principal a way to avoid this masking effects.
However, as the frequency response of the band-stop filter has to be included in the evaluation, the
whole measurement procedure may become somewhat complex.
For the resolution bandwidth, the recommended values should not exceed 30 kHz. Preferred values are
approx. 4 kHz. The measurement should be Noise-normalized to 4 kHz.
9.8 Receiver sensitivity/dynamic range for a Gaussian channel
Purpose For network planning purposes, the minimum and maximum input powers for normal operation of a
receiver have to be determined.
Interface Test signals are applied and measured at interface N; interfaces W or X are used for the monitoring of
BER before RS.
Method The minimum and maximum input power thresholds for QEF (Quasi Error Free) operation after the RS
decoder (i.e. BER < 2 × 10-4 before RS decoding) shall be measured. The dynamic range is the
difference between the measured values.
9.9 Equivalent Noise Degradation (END)
Purpose END is a measure of the implementation loss caused by the network or the equipment where the
reference is the ideal performance.
Interface W or X for BER measurement; N, P or S for noise injection
Method The END is obtained from the difference in dB of the C/N ratio needed to reach a BER of 2 × 10-4
× -4
ETSI
61 ETSI TR 101 290 V1.2.1 (2001-05)
before RS (outer) decoding, and the C/N ratio that would theoretically give a BER of 2 × 10-4 for a
Gaussian channel (see annex A of EN 300 744 [9]).
9.9.1 Equivalent Noise Floor (ENF)
Purpose ENF is a measure of the implementation loss caused by the transmitting equipment where the reference
is the ideal transmitter.
Interface M for noise power measurement, W or X for BER measurement; N, P or S for noise injection
Method The ENF is obtained from the measurement of additional noise needed to reach a BER of 2 × 10-4
before RS (outer) decoding, and the noise level that would theoretically give a BER of 2 × 10-4 for a
Gaussian channel (see annex A of EN 300 744 [9]) as described in clause B.12.
Note on END and ENF:
The impact of the DVB-T transmitter on the overall system performance, when a certain DVB-T mode is being received
by the reference receiver, via a Gaussian channel, is assessed by the measurement of the END.
The reference receiver is in the present document defined as a DVB-T receiver which require a C/N which is 3,0 dB
higher than the C/N figures indicated in EN 300 744 [9], on a Gaussian channel.
The END is in the present document defined to be the difference between required C/N, for a BER of 2 × 10-4 after
convolutional decoding on the reference receiver, using a real and an ideal DVB-T transmitter.
The END is not only a characteristic of the transmitter itself but is also dependent on the used DVB-T mode and on the
receiver implementation loss (this is why a fixed 3,0 dB receiver implementation loss is defined for the reference
receiver).
The END shall not exceed [0,5] dB and shall be independent of the selected guard interval. Depending on the
requirements of the network operator typical END values fall in the range [0,1-0,4] dB.
For the determination of the END value another parameter, the Equivalent Noise Floor ENF, can be used. As described
in clause B.12, this should result in an improved accuracy for the END.
As opposed to the END the ENF is relatively independent of the DVB-T mode used and on the receiver implementation
loss and can therefore be used to characterise the transmitter by itself. Depending on whether there is a need for
characterizing the DVB-T transmitter by itself, or whether there is a need to characterise its effect on a receiver, the
ENF can sometimes be used as an alternative to END as a performance parameter.
The influences of intermodulation and amplitude ripple are expected to dominate in practise in the performance
parameter END.
(The Group Delay response of a transmitter needs to be defined by network operators depending on the configuration
in use (channel combiners, output filters, etc).)
ETSI
62 ETSI TR 101 290 V1.2.1 (2001-05)
9.10 Linearity characterization (shoulder attenuation)
Purpose The "shoulder attenuation" can be used to characterize the linearity of an OFDM signal without
reference to a spectrum mask.
Interface M
Method Apply the following procedure on the measured RF spectrum of the transmitter output signal:
(a) Identify the maximum value of the spectrum by using a resolution bandwidth at approximately
10 times the carrier spacing.
(b) Place declined, straight lines connecting the measurement points at 300 kHz and 700 kHz from
each of the upper and lower edges of the spectrum. Draw additional lines parallel to these, so
that the highest spectrum value within the respective range lies on the line.
(c) Subtract the power value of the centre of the line (500 kHz away from the upper and lower
edge) from the maximum spectrum value of (a) and note the difference as the "shoulder
attenuation" at the upper and lower edge.
(d) Take the worst case value of the upper and lower results from (c) as the overall "shoulder
attenuation".
NOTE: For a quick overview the value at e.g. 500 kHz can be measured directly provided that
coherent interferers are not present.
9.11 Power efficiency
Purpose To compare the overall efficiency of DVB transmitters.
Interface M
Method Power efficiency is defined as the ratio of the DVB output power to the total power consumption of the
chain from TS input to the RF signal output including all necessary equipment for operation such as
blowers, transformers etc. (and is usually quoted in % terms). The operational channel and the
environmental conditions need to be specified.
9.12 Coherent interferer
Purpose To identify any coherent interferer which may influence the reliability of the I/Q analysis or the BER
measurements.
Interface N or P
Method The measurement is carried out with a spectrum analyser. The resolution bandwidth is reduced
stepwise so that the displayed level of the modulated carriers (and of the unmodulated pilots, due to the
influence of the guard interval) is reduced. The CWinterferer is not affected by this process and can be
identified after appropriate averaging of the trace.
9.13 BER vs. C/N ratio by variation of transmitter power
Purpose To evaluate the BER performance of a transmitter as the Carrier to Noise (C/N) ratio is varied, with the
measurement repeated for a range of mean transmitted output powers. This measurement can be used
to compare the performance of a transmitter with theory or with other transmitters.
Interface From F to U or from E to V
ETSI
63 ETSI TR 101 290 V1.2.1 (2001-05)
Method A Pseudo-Random Binary Sequence (PRBS) is injected at interface F (or E). The various C/N ratios
are established at the input of the test receiver by addition of Gaussian noise, and the BER of the
received PRBS is measured at point V (or U) using a BER TEST Set. The measurement is repeated for
a range of mean transmitted output power.
If the ability to generate a PRBS at interface F (or E) is included in the transmitting equipment for test
purposes, then it should be a 223-1 PRBS as defined by ITU-T Recommendation O.151 [12].
For the measurement of carrier and noise power, the system bandwidth is defined as n × fSPACING,
where n is the number of active carriers (e.g. 6 817 or 1 705 carriers in an 8 MHz channel) and
fSPACING is the frequency spacing of the OFDM carriers.
NOTE: Transmitter back-off is defined as the ratio of the rated pulsed peak power of the transmitter to the mean
power of the signal. The rated pulsed peak power is normally equivalent to the peak sync power of a
standard B, D, G, H, I or K RF signal.
9.14 BER vs. C/N ratio by variation of Gaussian noise power
Purpose To evaluate the BER performance of a receiver as the Carrier to Noise (C/N) ratio is varied by
changing the added Gaussian noise power. This measurement can be used to compare the performance
of a receiver with theory or with other receivers. For example to evaluate the influence of receiver
noise floor.
Interface From F to U or from E to V.
Method A Pseudo-Random Binary Sequence (PRBS) is injected at interface F (or E). Various C/N ratios are
established at the input of the receiver under test by addition of Gaussian noise and the BER of the
received PRBS is measured at point V (or U) using a BER test set.
A test transmitter should be able to generate the 223-1 PRBS as defined by
ITU-T Recommendation O.151 [12].
For the measurement of carrier and noise power, the system bandwidth is defined as n × fSPACING
where n is the number of active carriers i. e. 6 817 or 1 705 carriers and fSPACING is the frequency
spacing of the OFDM carriers.
NOTE: The bandwidth in an 8 MHz channel is approx. 7,61 MHz, in a 7 MHz channel system it is 6,66 MHz and
5,71 MHz in a 6 MHz channel.
9.15 BER before Viterbi (inner) decoder
Purpose This measurement gives an in-service indication of the un-coded performance of the transmitter,
channel and receiver.
Interface V.
Method The signal after Viterbi decoding in the test receiver is coded again using the same convolutional
coding scheme as in the transmitter in order to produce an estimate of the originally coded data stream.
This data stream is compared at bit-level with the signal which is available before Viterbi decoder.
The measurement should be based on at least several hundred bit errors.
ETSI
64 ETSI TR 101 290 V1.2.1 (2001-05)
Viterbi
Decoder
Delay
Convolutional
Coder
Outer
de-interleaver
Comparison
BER
V W X
Figure 9-4: BER measurement before Viterbi decoding
9.16 BER before RS (outer) decoder
Purpose The BER is the primary parameter which describes the quality of the digital transmission link.
Interface Wor X
Method The BER is defined as the ratio between erroneous bits and the total number of transmitted bits.
Two alternative methods are available; one for "Out of Service" and a second for "In Service" use. In
both cases, the measurement should only be done within the Link Available Time (LAT) as defined in
clause 6.2.
9.16.1 Out of Service
The basic principle of this measurement is to generate within the channel encoder a known, fixed,
repeating sequence of bits, essentially of a Pseudo-Random nature. In order to do this the data entering
the sync-inversion/randomization function is a continuous repetition of one fixed TS packet. This
sequence is defined as the null TS packet in ISO/IEC 13818-1 [1] with all data bytes set to 0x00; i.e.
the fixed packet is defined as the four byte sequence 0x47, 0x1F, 0xFF, 0x10, followed by 184 zero
bytes (0x00). Ideally this would be available as an encoding system option.
The apparently obvious alternative of injecting a PRBS in the transmitter at the output of the RS encoder is not used
because of the requirement to have sync bytes to ensure correct operation of the byte interleaver. Insertion after the
byte interleaver is not appropriate because it is not then directly comparable with the in-service measurement.
9.16.2 In Service
The basic assumption made in this measurement method is that the RS check bytes are computed for each link in the
transmission chain. Under normal operational circumstances, the RS decoder will correct all errors and produce an
error-free TS packet. If there are severe error-bursts, the RS decoding algorithm may be overloaded, and be unable to
correct the packet. In this case the transport_error_indicator bit shall be set, no other bits in the packet shall be changed,
and the 16 RS check bytes shall be recalculated accordingly before re-transmission on to another link. The BER
measured at any point in the transmission chain is then the BER for that particular link only.
The number of erroneous bits within a TS packet will be estimated by comparing the bit pattern of this TS packet before
and after RS decoding. If the measured value of BER exceeds 10-3 then the measurement should be regarded as
unreliable due to the limits of the RS decoding algorithm. Any TS packet that the RS decoder is unable to correct
should cause the calculation to be restarted.
9.17 BER after RS (outer) decoder (Bit error count)
Purpose To gain information about the pattern with which bit errors occur.
Interface Z
ETSI
65 ETSI TR 101 290 V1.2.1 (2001-05)
Method The same principle as used for the "Out of service" measurement of the "BER before the RS decoder"
described in clause 9.16.1, with the modification that the result is presented as an error count rather
than a ratio. The receiver only has to compare the received TS packets with the Null packets as defined
in clause A.1.2. This method is applicable for cases where the BER before RS decoder is lower than
approx. 10-3.
This can be used as one parameter for the estimation of the quality of the transmission link as it was
defined by the operator, or for localization of specific problems.
9.18 IQ signal analysis
9.18.1 Introduction
The IQ analysis can be applied on single carriers of the OFDM signal as well as on groups of carriers. If groups of
carriers are under consideration all received symbols of this group can be superimposed in order to get one common
constellation diagram. Since the scattered pilot carriers, the continual pilot carriers and the TPS carriers are transmitted
in a different modulation scheme it is recommended to exclude these carriers from the IQ analysis or apply a specific IQ
analysis.
Assuming:
- a constellation diagram of M symbol points and K carriers under consideration with
0 <K KMAX +1 and KMAX + 1 is the total number of active OFDM carriers (i.e. 1 705 or 6 817 carriers);
- a measurement sample of N data points, where N is sufficiently larger than M × K to deliver the wanted
measurement accuracy; and
- the co-ordinates of each received data point j being Ij + δIj, Qj + δQj where I and Q are the co-ordinates of the
ideal symbol point and δI and δQ are the offsets forming the error vector of the data point (as long as the
respective carrier is a "useful" one).
The following six parameters can be calculated, which give an in-depth analysis of different influences, all deteriorating
the signal.
Modulation Error Ratio (MER) and the related Error Vector Magnitude (EVM) are calculated from all N data points
without special pre-calculation for the data belonging to theM symbol points.
With the aim of separating individual influences from the received data, for each point i of the M symbol points the
mean distance di and the distribution σi can be calculated from those δIj, δQj belonging to the point i.
From the M values {d1, d2, ... dM} the influences/ parameters:
- Origin offset/ Carrier suppression (CS);
- Amplitude Imbalance; and
- Quadrature Error (QE)
(only for 2 k modes since the centre carrier needs to carry a complete constellation which is not the case in an 8k system
where the centre carrier is a continual pilot) can be extracted and removed from the di values, allowing to calculate the
Residual Target Error (RTE) with the same algorithm as the System Target Error (STE) from {d1, d2, ... dM}.
From the statistical distribution of the M clouds the parameters:
- Phase Jitter (PJ); and
- coherent interferer (if it is dominant)
may be extracted. The remaining clouds (after elimination of the above two influences) are assumed to be due to
Gaussian noise only and are the basis for calculation of the signal-to-noise ratio. The parameter may include - besides
noise - also some other disturbing effects, like small coherent interferers or residual errors from the channel correction.
ETSI
66 ETSI TR 101 290 V1.2.1 (2001-05)
When using the interfaces S or T filtering of the signal before the interface should be considered.
The parameters Origin offset/ Carrier suppression (CS), Amplitude Imbalance (AI) and Quadrature Error (QE) are
typical performance parameters of the modulator. The other parameters are also influenced by the transmission system
and the receiver/ demodulator.
It should be noted that the channel estimation/ channel correction mechanism can have an impact on the measurement
results. This is particularly true for measurements in the field or under simulated but realistic reception conditions.
For measurements taken at the output of a transmitter this impact of the channel estimation/ channel correction
mechanism is negligible.
For comparison of measurement results, information on the character of the channel estimation/ channel correction
mechanism should be provided.
9.18.2 Modulation Error Ratio (MER)
Purpose To provide a single "figure of merit" analysis of the K carriers.
Interface S, T and H
Method The carrier frequency of the OFDM signal and the symbol timing are recovered. Origin offset of the
centre carrier (e.g. caused by residual carrier or DC offset), Quadrature Error (QE) and Amplitude
Imbalance are not corrected.
A time record of N received symbol co-ordinate pairs ( ) I j Q j
~
,
~ is captured.
For each received symbol, a decision is made as to which symbol was transmitted. The error vector is
defined as the distance from the ideal position of the chosen symbol (the centre of the decision box) to
the actual position of the received symbol.
This distance can be expressed as a vector ( ) δ I j ,δQ j .
The sum of the squares of the magnitudes of the ideal symbol vectors is divided by the sum of the
squares of the magnitudes of the symbol error vectors. The result, expressed as a power ratio in dB, is
defined as the MER.
( )
( )
dB
I Q
I Q
MER
N
j
j j
N
j
j j










+
+
= ×


=
=
1
2 2
1
2 2
10 log10
δ δ
It should be reconsider that MER is just one way of computing a "figure of merit" for a vector
modulated signal. Another "figure of merit" calculation is Error Vector Magnitude (EVM) defined in
annex C of the present document. It is also shown in annex C that MER and EVM are closely related
and that one can generally be computed from the other.
MER is the preferred first choice for various reasons itemized in annex C of the present document.
9.18.3 System Target Error (STE)
Purpose The displacement of the centres of the clouds in a constellation diagram from their ideal symbol point
reduces the noise immunity of the system and indicates the presence of special kinds of distortions such
as Amplitude Imbalance and Quadrature Error (QE). STE gives a global indication about the overall
distortion present on the raw data received by the system.
Interface S and T.
Method For each of theM symbol points in a constellation diagram compute the distance di between the
theoretical symbol point and the point corresponding to the mean of the cloud of this particular symbol
ETSI
67 ETSI TR 101 290 V1.2.1 (2001-05)
point. This quantity ( di ) is called Target Error Vector (TEV) and is shown in figure 9-5.
I
Q di
ith point
Figure 9-5: Definition of Target Error Vector (TEV)
From the magnitude of the M Target Error Vectors (TEV) calculate the mean value and the standard
deviation (normalized to Srms , defined as the RMS amplitude value of the points in the constellation),
obtaining the System Target Error Mean (STEM) and the System Target Error Deviation (STED) as
follows:
TEV = di = (δIi ,δQi ) for all j = 1,2,...Ns data points belonging to the sub-symbol i;
with
=
=
s
s
N
j
Ii N I j
1
1 δ δ and
=
=
s
s
N
j
Qi N Qj
1
δ 1 δ
( )
=
= +
N
j
rms I j Qj
N
S
1
1 2 2
=
×
=
M
i
i
rms
d
M S
STEM
1
1
2
2
1
2
STEM
M S
d
STED
rms
M
i
i
−
×
=
=
9.18.4 Carrier Suppression (CS)
Purpose A residual carrier is an unwanted coherent signal added to the centre carrier of the OFDM signal. It
may have been produced by dc offset voltages of the modulating I and/or Q signal or by crosstalk from
the modulating carrier within the modulator.
Interface S and T.
ETSI
68 ETSI TR 101 290 V1.2.1 (2001-05)
Method Search for systematic deviations of all constellation points of the centre carrier and isolate the residual
carrier. Calculate the Carrier Suppression (CS) from the formula:






= ×
RC
sig
P
P
CS 10 log10
where PRC is the power of the residual carrier and Psig is the power of the centre carrier of the OFDM
signal (without residual carrier).
NOTE: Not applicable for 8k modes (see 9.18.1).
9.18.5 Amplitude Imbalance (AI)
Purpose To separate the QAM distortions resulting from Amplitude Imbalance (AI) of the I and Q signal from
all other kind of distortions.
Interface S and T.
Method Calculate the I and Q gain values vI and vQ from all points in a constellation diagram eliminating all
other influences.
Calculate Amplitude Imbalance (AI) from vI and vQ.
NOTE 1: Since the allocation of I and Q to the axis in the complex plane is unambiguous for a
DVB-T signal, the parameter AI can convey the information which component
dominates. Therefore, this definition differs slightly from the one given in 6.9.5.
ν
ν
Q I
I
Q
I Q
Q
I
v v
v
v v
v
if
if
AI
> ×






−
× ≥






−
=
1 100 %
1 100 % {
( )
( )
( )
( )
( ) ( ) i I i Q i
N
j
i Q j
M
i i
i i Q
Q
N
j
i I j
M
i i
i i I
I
d d d
Q
N
d
Q
Q d
M
I
N
d
I
I d
M
+ =
=
+
=
=
+
=




=
=
=
=
(Q- component of d asgiveninsubclause9.18.3)
1
1
(I - component of d asgiveninsubclause9.18.3)
1
1
i
1
1
i
1
1
δ
ν
δ
ν
NOTE 2: Not applicable for 8k modes (see 9.18.1).
ETSI
69 ETSI TR 101 290 V1.2.1 (2001-05)
9.18.6 Quadrature Error (QE)
Purpose The phases of the two carriers feeding the I and Q modulators have to be orthogonal. If their phase
difference is not 90 a typical distortion of the constellation diagram results.
It is assumed that the value derived from the centre carrier is representative for the whole signal.
Interface S and T.
Method Search for the constellation diagram error shown in figure9-6 and calculate the value of the phase
difference Δϕ = ϕ1 - ϕ2 after having eliminated all other influences and convert this into degrees:
= ° ×ϕ( −ϕ ) [°]
π 1 2
180
QE
I
Q
Decision Boundary
Signal Point
Decision Boundary Box
1 ϕ
2 ϕ
90°+QE
Figure 9-6: Distortion of constellation diagram resulting from I/Q
Quadrature Error (QE)
NOTE: Not applicable for 8k modes (see 9.18.1).
9.18.7 Phase Jitter (PJ)
Purpose The PJ of an oscillator is due to fluctuations of its phase or frequency. Using such an oscillator to
modulate a digital signal results in a sampling uncertainty in the receiver, because the carrier
regeneration cannot follow the phase fluctuations.
The signal points are arranged along a curved line crossing the centre of each decision boundary box as
shown in figure 9-7 for the four "Corner Decision Boundary Boxes".
ETSI
70 ETSI TR 101 290 V1.2.1 (2001-05)
Q
I
"Corner Decision Boundary Box"
for calculation of the Phase Jitter
Arc section through a
Figure 9-7: Position of "Arc section" in the constellation diagram to define PJ
(example: 64-QAM)
Interface S and T.
Method Phase Jitter can be calculated theoretically using the following algorithm:
1) Calculate the angle between the I-axis of the constellation and the vector to the received symbol
(I rcvd ,Qrcvd ) :
rcvd
rcvd
I
Q
φ1 = arctan
2) Calculate the angle between the I-axis of the constellation and the vector to the corresponding ideal
symbol (Iideal ,Qideal ) :
ideal
ideal
I
Q
φ 2 = arctan
Phi 2 instead of Phi 1
3) Calculate the error angle:
φ E =φ1 −φ2
4) From these N error angles calculate the RMS phase jitter:

= =






= −
N
i
N
i
N Ei N Ei
PJ
1
2
1
2
1 2 1 φ φ
However, the following method may be more practical:
The first approximation of the "Arc Section" of a "Corner Decision Boundary Box" is a straight line
parallel to the diagonal of the "Decision Boundary Box". Additionally the curvature of the Phase Jitter
(PJ) trace has to be taken into account when calculating the standard deviation of the PJ. The mean
value of the PJ is calculated in degrees.
( )






× − ×
= ° ×
M d
PJ PJ
2 1
arcsin
180 σ
π
[°]
ETSI
71 ETSI TR 101 290 V1.2.1 (2001-05)
where M = Order of QAM
and 2d = Distance between two successive boundary lines
Within the argument of the arc sine function, the standard deviation of the Phase Jitter is referenced to
the distance from the centre of the "Corner Decision Boundary Box" to the centre point of the QAM
signal.
9.19 Overall signal delay
Purpose To measure and adjust the signal delay of an OFDM transmitter to a given value so that the transmitters
in an SFN can be synchronized.
Interface A, M.
Method (a) The total delay between the MPEG TS input of the transmitter under test and the MPEG TS output
of a test receiver is established by measuring the time delay required to match the input and output data
patterns. If the delay of the test receiver is known then the transmitter signal delay can be derived.
Alternatively, the delay of the test receiver could be expressed relative to the delay of a reference
receiver. This would avoid the need to measure the absolute delay of any receiver.
(b) A more direct method may be to define a transmitter test mode in which the occurrence of a Megaframe
Initialization Packet (MIP) at the MPEG TS input causes a trigger pulse (see TS 101 191 [14]).
The trigger pulse is made available for connection to an oscilloscope and also used to "arm" the
modulator. At the start of the next mega-frame the modulator transmits a null symbol (or a defined
pulse in the time domain) rather than the normal data. The delay between the trigger pulse and the RF
null (or pulse) ismeasured.
(c) The delay of a transmitter could be expressed relative to the delay of a reference transmitter. For the
measurement a reduced amplitude sample is taken from both transmitters and adjusted to have similar
level (< 3 dB difference), the samples are combined in a RF linear adder and the output is fed to a
spectrum analyser. Typically the spectrum formed will have lobes due to the difference of delays in the
two transmitters. The inverse of the frequency width of the lobes represents the relative delay between
the transmitters.
Two drawbacks has to be taken in account:
1) the delay is absolute, that is, it gives no indication of which transmitter has the longer delay;
2) the accuracy is related to the ability of identifying the minimal values of the lobes and the
accuracy of the measurement.
NOTE 1: The delay of a transmitter may be considered as the addition of various parts including
the physical delays of the analogue part of the OFDM signal, including the path length to
the antenna. Also the buffers used for signal conditioning (TS bitrate adaptation to the
sampling frequency of the transmitter) and other intermediate buffers in the OFDM
spectrum calculation (IFFT) may differ from manufacturer to manufacturer.
NOTE 2: In cases of single frequency networks, the SFN adapter at the transmitter site may be
considered as integral part of the modulator transmitter. It may calculate the delay, from
the value of the STS (Synchronisation Time Stamp) to the 1 pps used as reference, in
different way from manufacturer to manufacturer and add differences in the delays that
have to be included in the measurement result.
It is recommended to use a test Transport Stream with embedded MIP data, and real-time calculation of
the STS.
See clause E.16 for test set-up, measurement description and example of results.
ETSI
72 ETSI TR 101 290 V1.2.1 (2001-05)
9.20 SFN synchronization
9.20.1 MIP_timing_error
Purpose A necessary precondition for SFN synchronization is that the Synchronization Time Stamp (STS)
values inserted in the Mega-frame Initialization Packet (MIP) are correct. This test checks that
successive STS values are self-consistent.
See TS 101 191 [14] .
Interface A, Z
(especially Transport Stream between the "SFN adapter" and "SYNC system" as defined in [14]).
Method Locate the MIP in three successive mega-frames numbered M, M+1 and M+2. Extract the
synchronization_time_stamp field from each MIP (STSM, STSM+1 and STSM+2).
In general, the difference between any two consecutive STS values will be the duration of one megaframe
minus some multiple (including zero) of the time between GPS pulses. Even without knowing
the precise duration of the mega-frame, we know that the duration is constant and can say that:
STSM+2 - STSM+1 = STSM+1 - STSM+ nT
where T is 1s and n is any integer.
Calculate nT from the above formula and check it is an integral number of seconds to within a user
defined accuracy.
This test can be performed continually on each successive set of 3 mega-frames, {M+1, M+2, M+3},
{M+2, M+3, M+4} etc. The test result must be discared if the mega-frame size changes over the set of
three mega-frames.
NOTE: The mega-frame size changes, for example, with the change of the DVB-T transmission
mode. This would normally result in a resynchronization.
NOTE: The following diagram is an illustration of the timing relationship between mega-frames and the GPS one
second pulses. This shows how the synchronization_time_stamp (STS) is calculated.
Consider STSM+1 and STSM+2. In this case it is quite clear that:
STSM+2 - STSM+1 = duration of one mega-frame
In the case of STSM and STSM+1, a 1s pulse has passed by and the equivalent equation is:
(STSM+1 + 1) - STSM = duration of one mega-frame
ETSI
73 ETSI TR 101 290 V1.2.1 (2001-05)
GPS 1s
pulses
Megaframe
M
I
P
M
I
P
M
I
P
M
I
P
M
I
P
STS
values
1s 2s 3s 4s
STSM
STSM+1
STSM+2
STSM+3
STSM+4
M M+1 M+2 M+3 M+4
Figure 9-8: Megaframe/ GPS pulse timing relationship
9.20.2 MIP_structure_error
Purpose This test verifies that the syntax of the MIP complies with the specification in
TS 101 191 [14].
Interface A, Z
Method For each transport packet carried on PID 0x15 in the transport stream, the following
checks are performed:
The transport_packet_header shall comply with TS 101 191 [14] clause 6, table 1, and
ISO/IEC 13818-1 [1] clause 2.4.3.2 tables 2 and 3.
All length fields must be consistent to provide a proper length packet. This includes
section_length (which also must not exceed 182), individual_addressing_length (which
must match the length of the loops for each transmitter), function_loop_length (which
must match the sum of the size of each of the functions), function_length (which must
match the proper length of the function based upon the function tag).
The synchronization_time_stamp and the maximum_delay must be in the range of 0x0
to 0x98967F.
The CRC_32 field must match the CRC calculated for the MIP data.
9.20.3 MIP_presence_error
Purpose This test verifies that the MIP is inserted into the transport stream only once per megaframe.
Interface A, Z
ETSI
74 ETSI TR 101 290 V1.2.1 (2001-05)
Method The following checks are performed:
Extra MIP – For every MIPN (where N > 1), signal an error if it arrives within the
number of packets indicated by the pointer field of MIPN-1.
Missing MIP - For each MIP received, calculate the mega-frame size from the
parameters in the tps_mip. The latest two values of the mega-frame size are stored.
After every MIPN is received (where N > 1), signal an error if a MIPN+1 is not received
before K + R packets are received after MIPN, where K is the pointer value of MIPN
and R is mega-frame size in packets from the previous MIPN-1.
9.20.4 MIP_pointer_error
Purpose The MIP insertion can be at any location in the mega-frame. If the insertion is periodic
as defined in the MIP, the MIP location in the mega-frame is constant over time. The
MIP can be used to determine the mega-frame size and where each mega-frame starts
and ends in the transport stream thanks to the pointer field verified by this test.
Interface A, Z
Method For each MIP received, calculate the mega-frame size from the parameters in the
tps_mip. The latest three values of the mega-frame size are stored. For everyMIPN that
is received (where N > 2), signal an error if the pointer value (PN) ofMIPN does not
hold in the following equation:
PN = PN-1+ MFN-2 - (iN - iN-1)
Where MFN-2 is the size of the Nth mega-frame in packets but is calculated from
MIPN-2, and iN is the packet index for MIPN.
9.20.5 MIP_periodicity_error
Purpose In the case of a periodic MIP insertion (as defined in TS 101 191 [14] clauses 5 and 6),
the pointer value shall remain constant, as well as the number of packets between each
MIP.
Interface A, Z
Method The following checks are performed:
Compare the current pointer field in MIPN with the pointer field in the MIPN-1. It is an
error if they are different, unless the mega-frame size changed between N and N-1.
The number of packets between each MIP (iN - iN-1) should also be constant unless the
mega-frame size changes.
ETSI
75 ETSI TR 101 290 V1.2.1 (2001-05)
9.20.6 MIP_ts_rate_error
Purpose In a SFN network the modulator settings are transmitted by the tps_mip (see TS 101 191 [14]
clause 6, table 3). These settings determine the transmission mode and in this way the bit rate
of the Transport Stream.
This test verifies that the actual Transport Stream data rate is consitent with the DVB-T mode
defined by the tps_mip.
Interface A, Z
Method For each MIP received, calculate the data rate of the transmission mode - given by tps_mip
setting and compare it with the actual data rate of the Transport Stream. Signal an error if the
following equation is correct:
Max_deviation ≤
| TS_data_rate - [(IFFT_clock_freq × tpl /204 × c × m × (uc/tc) )/(1 + g)] |
Where:
• Max_deviation e.g. 10 kb/s; maximum deviation between actual TS_data_rate
and data rate of the tranmission mode
given by tps_mip.
The value results from the smallest difference of
TS data rates which can be determined by two correct
tps_mip settings for different modes.
• TS_data_rate actual data rate of the Transport Stream
measured by a test instrument according to clause 5.3.3.2.
• IFFT_clock_freq 64/7 MHz (for 8 MHz channel bandwidth),
64/8 MHz (for 7 MHz channel bandwidth)
48/7 MHz (for 6 MHz channel bandwidth)
given by tps_mip P12 and P13
• tpl transport packet length 188 or 204 byte
• c coderate½, 2/3, ¾, 5/6 or 7/8
given by tps_mip P5,P6 and P7
• m 2 (for QPSK), 4 (for 16 QAM) or 6 (for 64 QAM)
given by tps_mip P0 and P1
• uc useful_carriers 1512 (for 2k), 6 048 (for 8k)
given by tps_mip P10, P11 (see note)
• tc total_carriers 2048 (for 2k), 8 192 (for 8k)
given by tps_mip P10, P11 (see note)
• g guard interval ¼, 1/8, 1/16 or 1/32
given by tps_mip P8, P9
NOTE: The term (uc/tc) can be replaced by a constant value since uc2k/tc2k = uc8k/tc8k.
ETSI
76 ETSI TR 101 290 V1.2.1 (2001-05)
9.21 System Error Performance
Purpose: The System Error Performance describes the performance of the digital transmission from the
input of the MPEG-2 TS signal into the DVB Baseline system to the MPEG-2 TS output of this
Baseline system.
Interfaces: A, Z,
M: with reference receiver (e.g. Transmitter measurement).
N: with reference receiver (e.g. coverage measurements).
Method: The measurement of System Error Performance is based on a subset of the error events defined in
clause 5.4:
- Errored Second (ES) or Errored Time Interval (ETI),
- Severely Errored Second (SES) or Severely Errored Time Interval (SETI).
The used time interval T for identification of these events depends on the aim of the measurement.
Time intervals longer or shorter than 1 second may be considered appropriate in certain
circumstances.
Evaluation of Error Performance Parameters
Error performance should only be evaluated whilst the transmission is in the available state (see
also 6.1).
To evaluate error performance parameters from events, a certain measurement interval (MI) has to
be used. This measurement interval depends on the specific aim of the measurement. Possible
measurement intervals corresponding to special applications are proposed in table 9.4.
In general the error performance is the ratio of number of true events to the total number of time
intervals T during the measurement interval.
Consequently derived performance parameters are:
- Errored Second Ratio (ESR) or Errored Time Interval Ratio (ETIR);
- Severely Errored Second Ratio (SESR) or Severely Errored Time Interval Ratio (SETIR).
Table 9.4: Examples of Measurement Intervals MI
Length of Measurement
Interval (MI)
Application
5 s - applicable for analysis of mobile reception
20 s - Coverage Check
- recommended minimum measurement interval for receiver comparison
5 minutes - possible resolution for 1 hour analysis.
1 hour - possible resolution for daily fluctuations analysis
10 Recommendations for the measurement of delays in
DVB systems
10.1 Introduction
For the measurement of the various types of delays which occur in a DVB system, including the encoder and decoder
for video and audio, the following parameters are defined:
- Overall delay;
- End-to-end encoder delay;
ETSI
77 ETSI TR 101 290 V1.2.1 (2001-05)
- Total decoder delay;
- Relative audio/ video delay (i.e. the difference of the overall delay for the video and the audio paths).
Encoding Presentation
Transmission
VBV VBV
t0 t2 t3
End to end encoding delay
t1
Decoding delay
PTS
ENCODER DECODER
Overall delay
Total decoder delay
Figure 10-1: Definition of delay parameters
NOTE: Tests on the overall delay of 4:2:0 codecs showed that the difference between the overall delay and the
end-to-end encoder delay is relatively small.
Measurements which included a SDI signal generator at the input of a MPEG encoder (working in 4:2:0 format) and the
PAL encoder incorporated in the MPEG decoder, showed values of 40 ms or 60 ms for the difference between the
overall delay and the calculated end-to-end encoder delay. The variation of 20 ms resulted from ambiguities related to
the point in time at which encoders and decoders were switched on, and were probably related to constraints of the PAL
encoder. It can be concluded that the difference between overall delay and end-to-end encoder delay will be 40 ms for a
SDI output, and the use of a PAL encoder may add 20 ms to this value.
The same results were obtained for various combinations of encoders and decoders from two different manufacturers.
In all cases the results were independent of the picture contents.
The proposals in this clause are described in such a way that mainly laboratory tests are considered (i.e. all pieces of
equipment are on the same site). This gives the tests the character of benchmark testing.
The relative audio/ video delay should also be checked to avoid potential problems. Especially in contribution and
production, it is advisable to measure this parameter. It may also be of interest for acceptance tests of encoders.
10.2 Technical description of the measurements
10.2.1 Definition of input signal
To ensure reliable detection at the Transport Stream layer, it is proposed to reflect the macroblock structure within the
active picture area such that a block of white lines starts at the second row of macroblocks, i.e lines 39 to 54 for 625
systems, and covers at least one row of macroblocks. It is recommended that the block of white lines is present for four
consecutive frames every 5 s.
ETSI
78 ETSI TR 101 290 V1.2.1 (2001-05)
10.2.2 Overall delay and end-to-end encoder delay
Encoder Decoder
Detector
TS
Recorder / Analyser
SDI Mux TS SDI
Detector
Figure 10-2: Measurement system description
The MPEG2 video encoder/multiplexer processes the SDI input signal defined in clause 2.1 to deliver a Transport
Stream output. The detector located at the input to the encoder/multiplexer is used to recognize either the video or audio
transition within the SDI input sequence and produces a signal to trigger recording of the transport stream by the TS
recorder analyser. An identical detector is placed at the SDI output of the decoder. This provides a trigger to the TS
recorder/analyser when the video or audio transition is decoded halting the recording.
This technique enables the measurement of two parameters:
- the overall delay;
- the end to end encoder delay;
(and the decoder delay after VBV buffer which equals the difference between overall delay and end-to-end
encoder delay)
10.2.2.1 Measurement of overall delay
The overall delay can simply be determined by measuring the time between the trigger produced by the detector located
at the input to the system and the trigger produced by the detector at the output of the decoder. The overall delay can be
measured for either video or audio depending on the nature of the detector. The accuracy of this measurement should be
±1 ms.
An alternative method makes use of the available audio path as a reference signal.
This procedure is based on use of equipment that is currently available, and operates with a special audio and video test
timing sequence. It comprises an audio test tone and a video signal that are gated synchronously with a period of 5 s.
allowing ±2,5 s. of audio-to-video delay measurement with an accuracy of 1 ms.
The audio tone consists of a sinewave with frequency selectable between 1 kHz and 10 kHz and levels selectable from
-20 dBu through +20 dBu.
The video signal currently comprises a black to white luminance transition on line 45 for the 525 lines format and on
line 38 for the 625 line format. In order to provide compatibility with the measurement equipment proposed for end to
end encoder delay measurement, and to ensure reliable detection at the Transport Stream layer, it is proposed that this is
modified to reflect the macroblock structure within the active picture area such that the block of white lines covers the
second row of macroblocks, i.e lines 39 to 54 for 625 systems.
Generators are available for analog and SDI formats with embedded audio.
ETSI
79 ETSI TR 101 290 V1.2.1 (2001-05)
Video/Audio
Generator
SDI/PAL/NTSC
Video
Encoder
Mux Decoder
Audio/Video
Delay Analyser
Audio Reference
Figure 10-3: Test set-up of overall video delay
The test set up for measurement of overall video delay is shown above. Note that the audio signal is fed directly to the
measurement set to act as a timing reference. Instruments available today provide a direct display of Audio–Video
Delay.
(Note that measurement of absolute audio delay can be also performed by using video as a reference).
10.2.2.2 Measurement of end to end encoder delay
The TS recorder/analyser begins recording, or analysing, when triggered by the detector located at the input to the
system and continues to record, or analyse, at least until the video or audio transition appears in the transport stream,
this can be ensured by only stopping the recording after the detector at the output of the decoder has generated a trigger.
The recorder analyser must locate the access unit where either the video or audio transition occurs. The latency time of
the encoder/multiplexer is then obtained by deducing the time between when the transition occurred at the input to the
system (the input trigger) and the time when the transition occurs in the transport stream (tlatency).
The end to end encoder delay also includes the buffer delay introduced by an ideal T-STD buffer model. This can be
calculated by analysing the actual, or interpolated, PTS of the transition access unit and the interpolated PCR at this
time. The difference between the PTS of the transition access unit and the interpolated PCR at this time gives a good
approximation of the decoder buffer delay in the end to end 'encoding' delay (tbuffer_delay).
The end to end encoder delay is the addition of the encoder latency and the buffer delay.
tend_to_end_encoder_delay = tlatency + tbuffer_delay
For reasons of comparison, it is recommended that the values of end-to-end encoder delay are measured for the
following combinations of profiles, bit rates and GOP structures:
MPEG2 coding
profile
Bit Rate Ru
(after Mux)
( Mbit/s) (6)
End to end encoder delay
( ms)
I only Low Delay IBP (4)
MP@ML (5) 4,6078 (1)
8,4480 (2)
4.2.2@ML(5) 21,5030 (3)
(1) With a minimum Elementary Stream (ES) video rate of 3 Mbit/s.
(2) With a minimumES video rate of 7 Mbit/s.
(3) With a minimumES video rate of 20 Mbit/s.
(4) With a GOP length of 12.
(5) Resolution is 720 x 576 for video frame rate of 25 Hz and 720 x 480 for video frame rate of 29,97 Hz.
(6) Considering 188 bytes format.
10.2.2.3 Total decoder delay measurement.
The total delay introduced by the decoder from TS input to SDI output may be measured by determining the time
between the TS packet which contained the access unit where either the video or audio transition occurs and the trigger
produced by the detector at the output of the decoder.
ETSI
80 ETSI TR 101 290 V1.2.1 (2001-05)
10.2.2.4 Measurement of Relative Audio/Video delay - Lip Sync
The test signals described earlier for measurement of overall delay may also be used for measurement of relative
audio/video delay - lip sync.
Video/Audio
Generator
SDI/PAL/NTSC
Video/Audio
Encoder
Mux Decoder
Audio/Video
Delay Analyser
Figure 10-4: Test set-up for relative audio/ video delay
The test set up is shown in the above diagram. In this case, both audio and video signals are fed through the codec path.
Relative Audio/Video Delay can be displayed directly.
The test procedure should ensure that the measurement is stable. It is also recommended that the power for the decoder
should be cycled to show repeatability.
ETSI
81 ETSI TR 101 290 V1.2.1 (2001-05)
Annex A (informative):
General measurement methods
A.1 Introduction
It is recommended that manufacturers add the test mode described in this annex to certain professional grade cable and
satellite broadcast equipment. This recommendation is relevant to equipment that implements the channel encoding
schemes defined in EN 300 429 [6] (cable) and EN 300 421 [5] (satellite).
The purpose of the recommended test mode is to simplify out of service testing of systems and system components by
making the channel encoder able to generate a known, fixed, repeating bit sequence of an essentially pseudo-random
nature.
The central requirement is that when the channel encoder is in the test mode, the data entering the sync
inversion/randomization function is a continuous repetition of one fixed TS packet. The fixed packet is defined as the
four byte sequence 0x47, 0x1f, 0xff, 0x10, followed by 184 zero bytes (0 x 00). This form of data is a refinement of the
null TS packet definition in ISO/IEC 13818-1 [1].
A.2 Null packet definition
This clause summarizes the null packet definition from ISO/IEC 13818-1 [1] and then describes how the definition has
been extended for the purpose of the recommended test mode.
ISO/IEC 13818-1 [1] defines a null TS packet for the purposes of data rate stuffing.
Table A.1 shows the structure of a null TS packet using the method of describing bit stream syntax defined in
clause 2.4.3.3. of ISO/IEC 13818-1 [1].
This description is derived from tables 2-3 Transport Header (TH) in ISO/IEC 13818-1 [1]. The abbreviation "bslbf"
means "bit string, left bit first", and "uimsbf" means "unsigned integer, most significant bit first".
The column titled "Value", gives the bit sequence for the recommended null packet.
A null packet is defined by ISO/IEC 13818-1 [1] as having:
- payload_unit_start_indicator = "0";
- PID = 0x1FFF;
- transport_scrambling_control = "00";
- adaptation_field_control value = "01". This corresponds to the case "no adaptation field, payload only".
The remaining fields in the null packet that shall be defined for testing purposes are:
- transport_error_indicator which is "0" unless the packet is corrupted. For testing purposes this bit is defined as
"0" when the packet is generated;
- transport_priority which is not defined by ISO/IEC 13818-1 [1] for a null packet. For testing purposes this bit
is defined as "0";
- continuity_counter which ISO/IEC 13818-1 [1] states is undefined for a null packet. For testing purposes this
bit field is defined as "0000";
- data_byte which ISO/IEC 13818-1 [1] states may have any value in a null packet. For testing purposes this bit
field is defined as "00000000".
ETSI
82 ETSI TR 101 290 V1.2.1 (2001-05)
Table A.1: Null TS packet definition
Syntax No. of bits Identifier Value
null_transport_packet(){
sync_byte 8 bslbf "01000111"
transport_error_indicator 1 bslbf "0"
payload_unit_start_indicator 1 bslbf "0"
transport_priority 1 bslbf "0"
PID 13 uimsbf "1111111111111"
transport_scrambling_control 2 bslbf "00"
adaptation_field_control 2 bslbf "01"
continuity_counter 4 uimsbf "0000"
for (I = 0;i<N;i++) {
data_byte 8 bslbf "00000000"
}
}
A.3 Description of the procedure for "Estimated Noise
Margin" by applying statistical analysis on the
constellation data
Instead of adding real noise to the received signal this method uses statistical analysis and an iterative search algorithm
to estimate the added noise power to reach the critical BER.
1) Demodulate the signal to produce a statistically significant sequence of data records. Each record represents the
state of the demodulated I and Q components at a decision instant.
2) Compute the average noise power as the mean square of the error vectors and calculate the estimated Savg/Navg
ratio.
( )
( )










+
+
= ×


=
=
N
j
N j j
N
j
N j j
I Q
I Q
SNR
1
1 2 2
1
1 2 2
10 log10
σ σ
The σIj and σQj are the error vector co-ordinates which represent the offset from the co-ordinates of the centre (mean
value) of the actual received data for a specific constellation point, to the actual received data point j (see also
figure 6-2).
If only Gussian noise is present as an impairment the "centre (mean value) of the actual received data fora specific
constellation point" is identical to the ideal symbol point.
N is the number of data points in the measurement sample.
3) Compute the additional noise power Nstep required to degrade the computed SNR by a certain amount. The value
Nstep is usually determined by the iterative optimization procedure which is used.
4) For each data record in the sample compute the distances d from the true position of the signal at the decision
instant to each of the decision boundaries with adjacent cells. For each of the directions +I, -I, +Q, -Q that would
cause a symbol error, convert the distance to the decision boundary into the number of standard deviations (k) of
a normal distribution with a variance corresponding to the added noise power. The variance of the added noise
power is:
= Nstep σ 2
ETSI
83 ETSI TR 101 290 V1.2.1 (2001-05)
and the normalized standard deviation corresponding to the distance dI+ is for example:
+
+ =
I
I d
k
σ
5) Compute the probability QS of a symbol error for each distribution tail due to an erroneous state transition in
the relevant direction.
( )
∞






= −
k
s dx
x
Q k
2
exp
2
1 2
π
or
( )





=
2 2
1 k
Qs k erfc
6) Compute the number of bit errors that the erroneous state transition would cause and calculate the bit error
probability QB. One symbol error may result in more than one bit error for transitions across either the I or Q
axis. Sum the individual QB values and divide by the number of points in the sample to get the average
probability of a bit error.
7) Repeat the steps 4 to 6 for incremental values of noise power until the critical BER is found and calculate the
noise margin:
Noise Margin






= × +
avg
added
N
N
(dB) 10 log10 1
A.4 Set-up for RF phase noise measurements using a
spectrum analyser
The noise performance of the carrier can be characterized as the ratio of the measured power in one noise sideband
component, on a per hertz of bandwidth spectral density basis, to the total signal power:
( )





= ×
power_of_total_signal
power_density(one_sideband,phase_only)
α fm 10 log10
in (dBc/Hz) and fm is the frequency distance away from the carrier.
For this measurement it is assumed that contributions from amplitude modulation to the noise spectrum are negligible
compared to those from frequency modulation and that ΔB, the measurement bandwidth, is much smaller than fm. A
spectrum analyser with a noise measurement option is able to measure the power within 1 Hz bandwidth. If this is not
available the resolution bandwidth should be as small as possible and the video bandwidth has to be 10 or 20 times
smaller in order to get sufficient averaging of the noise over time.
For example: carrier frequency: 36 MHz
fm= 10 kHz
ΔB = Equivalent Noise Bandwidth (ENB) of the resolution bandwidth filter: 270 Hz
video bandwidth: 10 Hz or 30 Hz
NOTE 1: Spectrum analysers typically use near Gaussian filters for the resolution bandwidth with a 20 % tolerance.
The Equivalent Noise Bandwidth (ENB) is equal to the bandwidth of the filter measured at -3,4 dB, (by
actually measuring the filter of the spectrum analyser, the 20 % tolerance factor is eliminated).
ETSI
84 ETSI TR 101 290 V1.2.1 (2001-05)
Then the following conversion to 1 Hz bandwidth can be applied:
( ) 10 log B
signal_power
noise_power_in_DB
log 10 10 10 Δ × −





α fm ≅ × + 2,5 dB in [dBc/Hz]
NOTE 2: The 2,5 term accounts for the correction of 1,05 dB due to narrowband envelope detection and the
1,45 dB due to the logarithmic amplifier.
Having measured α(fm) for various values of fm an estimation of equivalent peak phase deviation and frequency
deviation is possible by using sinusoidal analogy:
α(fm) ≅ 20 × log10(Δϕrms/ 2 ) in [dB/Hz]
with Δϕ in [rad/Hz]
The square root of the sum of all noise densities within the frequency range of interest will give the equivalent RMS
phase noise error vector in the I/Q plane.
An estimation can be done if the phase noise power slope may be approximated by the density function:
[W Hz]
f
Y a
b
1 =
with
10
slope[dB]_ per _ decade
b = (b > 0) and
a N f b = 0 × 1 where
( )




= 10
0
1
10
f
N
α
Then the total double-side-band phase noise power within the frequency range of interest (f1,f2) can be approximated
by:
( ) ( ) ( )






−
−
− − = = − −
2
1
1
2
1
1
1 1
1
1 2
2
f
f
b b f b f b
a
df
f
DSB Phase Noise a
For the normalized RMS error vector (carrier = 1) it follows:
RMS Quadrature Error Vector (b ) f (b ) f (b ) ph
a σ =






−
−
= − −1
2
1
1
1 1
1
2
Δϕσ ≅ arctanσ ph [rad] (for carrier = 1)
A.5 Amplitude, phase and impulse response of the
channel
The amplitude, phase and impulse response can be derived from the equalizer tap coefficients. The use of a good
equalizer that is designed to cope with the echo profile defined in clause B.14 is recommended to get accurate results in
case of high linear distortions.
The capabilities to derive the channel response from the equalizer tap coefficients depend on the structure of the
equalizer. Especially the channel response in the Nyquist slope of the signal can not be measured exactly with a
T-spaced equalizer.
ETSI
85 ETSI TR 101 290 V1.2.1 (2001-05)
A.6 Out of band emissions
The out of band emissions can be measured using a spectrum analyser. The resolution bandwidth shall be low enough to
detect peaks in the out of band spectrum. The video filter shall be at least 10 times lower than the resolution bandwidth
for sufficient averaging of the noise-like signal.
Figure A-1: Spectrum mask as defined in EN 300 429 [6]
ETSI
86 ETSI TR 101 290 V1.2.1 (2001-05)
Annex B (informative):
Examples for test set-ups for satellite and cable systems
Even if not demonstrated in the diagrams of this clause and also not mentioned in the explanations the receiver may be a
part of the measurement device. In this case all the interfaces defined in figure 4-2 are internal ones, where the
measurement device has access to.
B.1 System availability
See clause 6.1.
Because this measurement is based on the error_indicator_flag in the TS header set in any previous stage including the
last stage of the transmission chain the signal at interface Z shall be used.
Figure B-1: Test set-up for system availability
B.2 Link availability
See clause 6.2.
This measurement monitors the performance of an individual link. Therefore the RS information shall be created and be
correct at the start point of the link. The measurement set-up may rely on the overload information coming from the RS
decoder in the receiver at interface X or on the transport_error_indicator in the header of the TS packets at interface Z.
Figure B-2: Test set-up for link availability
B.3 BER before RS
See clause 6.3.
The measurement can be done as out service measurement or as in service measurement. In both cases the measurement
time is an important parameter. This parameter should be selectable within a wide range by the user. Preferably the
measurement should display the BER as a function of measurement time.
ETSI
87 ETSI TR 101 290 V1.2.1 (2001-05)
B.3.1 Out of service measurement
See clause 6.3.1.
When the BER is measured out of service Null packets as defined in clause A.2 shall be created and transmitted to the
receiving site. At the receiving site the signal at the interfaceW is compared against the pre-calculated values. The time
window for the BER measurement should be selectable by the user.
Figure B-3: Test set-up for out of service BER measurement before RS decoding
B.3.2 In service measurement
See clause 6.3.2.
In this case no special signal shall be inserted in the transmitter. The measurement only relies on the results of the RS
decoder. The measurement can be done by using the signals at the interfaces W and Z.
Figure B-4: Test set-up for out of service BER measurement before RS decoding
B.4 Event error logging
See clause 6.4.
This measurement relies on information coming from different parts of the receiver like tuner, RS decoder or a
demultiplexer. Typically the receiver will be a part of the measurement device because it is not expected that all this
information will be available at a standard receiver.
Figure B-5: Test set-up for event error logging
ETSI
88 ETSI TR 101 290 V1.2.1 (2001-05)
B.5 Transmitter symbol clock jitter and accuracy
See clause 6.5.
This measurement requires a symbol clock output at the modulator. To this interface an appropriate frequency counter
and/or jitter and wander analyser can be connected.
Figure B-6: Test set-up for transmitter symbol clock measurement
B.6 RF/IF signal power
See clause 6.6.
The signal power can be measured directly at the interfaces N or P or by using a calibrated splitter. If needed an
appropriate filter should be used.
Figure B-7: Test set-up for RF/IF level measurement
B.7 Noise power
See clause 6.7.
Typically all the power present in a channel which is not part of the signal can be regarded as unwanted noise. It can be
produced from different origination and be of the form of random noise (thermal), pseudo-random (digitally modulated
interfering carriers) or periodic (Continuous Waves CW or narrowband interferences), the first two are called noncoherent
and the periodic ones are termed as coherent.
B.7.1 Out of service measurement
For doing this measurement the carrier shall be switched off. The measurements can be done at interface N (RF level)
or at interface P (IF level). Noise level can be measured with a spectrum analyser or any other appropriate device. If a
power metre is used the equivalent noise bandwidth should be taken into account. In this case of out-of-service
measurement, all different types of noise are measured simultaneously, and the measured result can be termed as
unwanted power.
ETSI
89 ETSI TR 101 290 V1.2.1 (2001-05)
Figure B-8: Test set-up or out-of-service noise level measurement
B.7.2 In service measurement
For the "in service measurement" eye diagrams or IQ constellation diagram derived from I and Q signals available at
interface T shall be employed. In this case of "in service measurement", it is possible to determine the type of the noise
by applying the I/Q signal analysis (see clause 6.9).
Figure B-9: Test set-up for in-service noise level measurement
B.8 BER after RS
See clause 6.8.
The set-up is equivalent to clause 6.3 BER before RS. The comparison is done after RS at interface Y.
B.9 I/Q signal analysis
See clause 6.9.
For this measurement eye diagrams or IQ constellation diagram derived from I and Q signals available at interface T
shall be employed.
Figure B-10: Test set-up for I/Q signal analysis
B.10 Service data rate measurement
The set-up is equivalent to B.1 The measurement is based on the TS only.
B.11 Noise margin
See clause 7.1.
Purpose To provide an indication of the reliability of the transmission channel (i.e. cable network), the noise
margin measurement is a more useful measure of system operating margin than a direct BER
ETSI
90 ETSI TR 101 290 V1.2.1 (2001-05)
margin measurement is a more useful measure of system operating margin than a direct BER
measurement due to the steepness of the BER curve.
Interface The reference interface for the noise injection is the RF interface (N, input of receiver). For
practical implementation, other interfaces can be used, provided equivalence to the described set-up
is ensured.
Test set-up Figure B-11 shows the recommended test set-up for the measurement of noise margin.
Figure B-11: Test Set-up for noise margin measurement
B.11.1 Recommended equipment
1 I/Q baseband signal source for 64 QAM;
S switch (to switch off modulation);
2 I/Q modulator;
3 RF generator (see clause B.11.2 below, remark 1) (level and frequency adjustable);
4 cable network (see clause B.11.2, remark 2);
5 noise source (flat within the required measurement range) (see clause B.11.2, remark 3);
6 adjustable attenuator in 0,1 dB (max. 0,5 dB) steps;
7, 8 directive couplers (see clause B.11.2, remark 4);
9 spectrum analyser;
10 reference receiver with good equalizer (see clause B.11.2, remark 5);
11 counter of BER.
B.11.2 Remarks and precautions
1) Adjust RF carrier level so that non-linear distortion (i.e. CW, CSO, CTB) has no impact to BER measurement.
2) Pay attention to the amplitude response of the noise spectrum. If it is not white Gaussian spectrum (flat
amplitude response) figure B-12 take care to measure:
ETSI
91 ETSI TR 101 290 V1.2.1 (2001-05)
a) If the effect produced by the thermal random noise is the wanted measurement, then take the measurement at
the lowest level found in the wanted band (P4 in figure B-12), because it is the closest approximation to the
random white thermal noise, then normalize the result to the full bandwidth of the channel, defined by the
symbol rate x(1 + α).
b) If the mean unwanted power is to be reported in the measurement, then integrate the spectrum with a suitable
spectrum analyser or use a power metre with the appropriate filter as per clause B.7.1.
Figure B-12: Amplitude response of the noise spectrum
3) If a noise source with broadband output spectrum is used, avoid any affect to BER measurement by non-linear
distortion due to an overload of the reference receiver's input amplifier stage.
4) Usual power splitters are allowed if sufficient matching at all ports is ensured for all measurement conditions
(i.e. high attenuation in adjustable attenuator).
5) Influence of linear distortion of the cable network to the BER measurement should be negligible.
B.11.3 Measurement procedure
Step 1:Add noise to modulated cable network output until BER is 10-4.
Step 2:Switch off modulation with (S);
Measure Noise power N1 (dBm) beside carrier (Δ f ≥ 0,5 MHz).
Step 3:Switch off noise source (5);
Measure Noise power N2 (dBm) beside carrier.
Step 4:Compute Noise Margin (NM):
NM = N1 − N2 (dB)
NOTE: Due to step 2, the measurement of noise margin is to be done under out of service conditions.
B.12 Equivalent Noise Degradation (END)
Figure B-13: Test set-up for END measurement
ETSI
92 ETSI TR 101 290 V1.2.1 (2001-05)
Procedure for the measurement of one point in the diagram:
1) Measure the power of the DVB signal with a power metre. If this is not possible due to signals in the
neighbouring channels, use a calibrated spectrum analyser.
2) Remove the wanted input signal and terminate the input.
3) Add noise to obtain the same level on the spectrum analyser. Now C/N = 0 dB.
4) Add the wanted input signal and increase the attenuation of the noise until a BER of 10-4 is measured. The
value, for which the attenuation was increased, is the C/N for the given BER.
5) END is the difference between the measured C/N and the theoretical value of C/N for a BER of 10-4.
Proposed settings for the spectrum analyser: RBW = 30 kHz, VBW < 300 Hz.
B.13 BER vs. Eb/N0
The BER versus Eb/N0 will be measured using the test set-up described above.
C/N measurements can be converted to Eb/N0 using the following formula:
N 10log (m)
Eb/N0 C = − 10
NOTE: For consideration of FEC overhead, see also 7.5, 8.2, G.5, G.6 and G.7.
B.14 Equalizer specification
High order modulations such as 64 QAM are very sensitive to distortions. The eye aperture is so small that any
perturbation can seriously disturb the reception of the signal. In the case of the DVB modulation formats, this problem
is increased by the low value of the roll-off factor (0,15). In a real network, if no special processing is carried out in the
receiver, the eyes appear completely closed, and no synchronization is possible. This is why all cable receivers,
professional or not, are equipped with equalizers.
Some of the most common impairments met on cable networks are echoes due to equipment impedance mismatching,
or filtering effects. These impairments appear as perturbations of the frequency response (or impulse response) of the
channel, and are corrected by the equalizer which is a form of adaptive filter. Equalizers are very efficient for linear
distortions, but cannot combat those of a non-linear nature. They combat fixed frequency interference, which is
equivalent to intermodulation products of analogue television signals. Equalizers have a large influence on the clock or
carrier recovery systems, since these can use equalized signals. Thus the overall behaviour of the receiver depends on
the performance of the equalizer.
Most of the measurements specified in the present document are carried out after equalization. The first reason is that
the signal is too impaired before equalization to obtain meaningful measurement results. Moreover, as most of the
distortion at that point would be removed in any practical receiver, such measurements may not be relevant. The
consequence of this is that measurement results are dependant on the equalizer response. This also means that
equipment with different equalizer architectures will have different performance characteristics. This situation is not
acceptable, and has led to the specification of the equalizer.
The specification of an equalizer is a difficult task, because there is a large number of types of equalizer, due to the
range of algorithms for the updating of coefficients, and the different filter architectures (time based, frequency based,
recursive or non-recursive). In addition, the performance of future equipment should not be limited by any specification
here. This is why a convenient solution is to specify the overall performance of the receiver as regards a perturbation
typically corrected by the equalizer, specifically - echoes.
The specification has to be defined so that the reference perturbation does not affect the measurement. We will then
define the minimum level of perturbation that the equalizer will have to correct. A solution is to set the minimum level
of an echo that will not degrade the equivalent noise degradation of the incoming signal of more then 1 dB. This
measurement is carried out for the worst case phase shift of the echo.
ETSI
93 ETSI TR 101 290 V1.2.1 (2001-05)
Figure B-14 gives a possible equalizer specification which is subject to verifications in real systems.
Figure B-14: Specification of an equalizer
In some cases, when the likely response of a consumer receiver to network signals is studied, it is appropriate to have an
equalizer in the measurement equipment whose performance is close to that of the consumer receiver.
B.15 BER before Viterbi decoding
This measurement shall be based on the I and Q signals at interface T. If an external measurement device is used the
signals at interfaces T and V are needed. The set-up is equivalent to figure B-9.
B.16 Receive BER vs. Eb/N0
The measurement is based on transmission of Null packets as defined in A.2. At the receiving site noise is added at one
of the interfaces N, P or R. The spectrum analyser is used for checking that the normal noise level is well below the
added noise. The measurement itself is done either within the receiver or at one of the interfaces T, V or Y depending
whether BER before Viterbi, after Viterbi or after RS shall be evaluated. In case of interface Y, RS decoding should be
deactivated in order to reduce the duration of the measurement.
ETSI
94 ETSI TR 101 290 V1.2.1 (2001-05)
Figure B-15: Test set-up for BER vs. Eb/N0 measurement
B.17 IF spectrum
The output of the modulator shall be directly connected to the spectrum analyser. In addition it is also possible to use a
(calibrated) splitter.
Figure B-16: Test set-up for IF spectrum measurement
ETSI
95 ETSI TR 101 290 V1.2.1 (2001-05)
Annex C (informative):
Measurement parameter definition
C.1 Definition of Vector Error Measures
Modulation Error Ratio (MER) is defined as:
( )
( )
( )
( )
dB
I Q
I Q
dB
I Q
I Q
MER
N
j
j j
N
j
j j
N
j
j j
N
j
j j










+
+
= ×










+
+
= ×




=
=
=
=
1
2 2
1
2 2
10
1
2 2
1
2 2
10 log10 20 log
δ δ δ δ
Error Vector Magnitude (EVM) is defined as:
( )
100%
1
2
max
1
2 2
×
+
=
=
S
I Q
N
EVM
N
j
j j
RMS
δ δ
Where I and Q are the ideal co-ordinates, δI and δQ are the errors in the received data points. N is the number of data
points in the measurement sample. Smax is the magnitude of the vector to the outermost state of the constellation.
C.2 Comparison between MER and EVM
To compare the two measures it is easier to write them both as simple ratios, clearly the use of decibels and percentages
is not central to the definition. Taking MER first, the simple voltage ratio (MERV) is:
( )
( )











+
+
=


=
=
N
j
j j
N
j
j j
V
I Q
I Q
MER
1
2 2
1
2 2
δ δ
and multiplying both numerator and denominator by N
1 gives:
( )
( )











+
+
=


=
=
N
j
j j
N
j
j j
V
I Q
N
I Q
N
MER
1
2 2
1
2 2
1
1
δ δ
ETSI
96 ETSI TR 101 290 V1.2.1 (2001-05)
Now looking at EVM as a simple voltage ratio (EVMV), we can write:
( )
max
1
1 2 2
S
I Q
N
EVM
N
j
j j
V
=
+
=
δ δ
EVM and MER are related such that:
( )
max
max
1
2 2
/
1
1
S S
S V
I Q
N
MER EVM rms
N
j
j j
V V = =
+
× =
=
or
MER V
EVM
V
V ×
= 1
If the peak to mean voltage ratio, V, is calculated over a large number of symbols (10 times the number of points in the
constellation is adequate if the modulation is random) and each symbol has the same probability of occurrence then it is
a constant for a given transmission system. The value tends to a limit which can be calculated by considering the peak
to mean of all the constellation points. Table A.2 lists the peak- to-mean voltage ratios for the DVB constellation sizes.
Table C.1: Peak-to-mean ratios for the DVB constellation sizes
QAM format Peak-to-mean voltage ratio
(V)
16 1 341
32 1 303
64 1 527
C.3 Conclusions regarding MER and EVM
MER and EVM measure essentially the same quantity and easy conversion is possible between the two measures if the
constellation is known. When expressed as simple voltage ratios MERV is equal to the reciprocal of the product of
EVMV and the peak-to-mean voltage ratio for the constellation.
MER is the preferred measurement for the following reasons:
- The sensitivity of the measurement, the typical magnitude of measured values, and the units of measurement
combine to give MER an immediate familiarity for those who have previous experience of C/N or SNR
measurement.
- MER can be regarded as a form of Signal-to-Noise ratio measurement that will give an accurate indication of a
receiver's ability to demodulate the signal, because it includes, not just Gaussian noise, but all other
uncorrectable impairments of the received constellation as well.
- If the only significant impairment present in the signal is Gaussian noise then MER and SNR are equivalent.
ETSI
97 ETSI TR 101 290 V1.2.1 (2001-05)
Annex D (informative):
Exact values of BER vs. Eb/N0 for DVB-C systems
Exact values of BER vs. Eb/N0 for DVB-C systems (see figure 7-2).
Table D.1: Exact values of BER vs. Eb/N0 for DVB-C systems
Eb/N0 Pb
10 0,025 48
10,5 0,020 72
11 0,016 46
11,5 0,012 74
12 0,009 582
12,5 0,006 981
13 0,004 909
13,5 0,003 319
14 0,002 147
14,5 0,001 323
15 0,000 771 6
15,5 0,000 423 5
16 0,000 217 1
16,5 0,000 103 1
17 4,499e-005
17,5 1,783e-005
18 6,351e-006
18,5 2,006e-006
19 5,537e-007
19,5 1,314e-007
20 2,634e-008
20,5 4,365e-009
21 5,846e-010
21,5 6,166e-011
22 4,974e-012
This assumes that the relationship between BER and Symbol Error Rate (SER) is given by the formula:
SER
m
BER = × 1
ETSI
98 ETSI TR 101 290 V1.2.1 (2001-05)
Annex E (informative):
Examples for the terrestrial system test set-ups
Due to the essential differences in the modulation method used for the terrestrial system some of the test methods are
different with respect to those used for cable and/or satellite.
Even if not demonstrated in the diagrams of this clause and also not mentioned in the explanations, the receiver may be
a part of the measurement device. In this case all the interfaces defined in figure 9-2 are internal ones, which the
measurement device has access to.
E.1 RF frequency accuracy
See clause 9.1.
DVB-T
Tx
Reference
Signal Source
Spectrum
Analyser
DUT
L, M
Figure E-1: RF frequency accuracy set-up
The measurement is to be done with a spectrum analyser. The signal can be picked up at interface L (IF) or M (RF),
eventually by means of an aerial, or at interface N, if the received signal can be maintained stable enough for the
measurement purposes, and applied to a spectrum analyser. Care should be taken at interfaces L or Mnot to overdrive
the maximum allowed input signal for the spectrum analyser.
E.1.1 Frequency measurements in DVB-T
A relatively easy conceptual model for creating an OFDM signal is by means of the Inverse Discrete Fourier Transform
(IDFT). This transform may be implemented by one of the several available algorithms called Fast Fourier Transform
(FFT) by virtue of their capability of saving computation time. The reverse process is called IFFT (Inverse Fast Fourier
Transform). Most of these algorithms are based on using an array of samples that has a length of a power of two.
For example, an array of 214 = 16 384 time domain samples can be processed to provide two arrays of 8 192 samples
representing the two, real and imaginary, array samples of the frequency domain by a direct FFT. The reverse applies
when change from frequency domain to time domain.
The 8k mode of DVB-T is defined to use 6 817 carriers, then it seems appropriate to use the above sized arrays of
8 192 (213) samples in the frequency domain, hence the name for the mode of 8k.
The 2k mode of DVB-T is defined to use 1 705 carriers, then it seems appropriate to use arrays of 2 048 (211) samples
in the frequency domain, hence the name for the mode of 2k.
The standard EN 300 744 [9] (DVB-T) defines each single symbol of OFDM as a sum of terms ranging from kmin to
kmax and those values are 0 to 6 816 for the 8k mode and 0 to 1 704 for the 2k mode. The central carriers have indexes
of 3 408 and 852 respectively.
ETSI
99 ETSI TR 101 290 V1.2.1 (2001-05)
Clause D.2 of EN 300 744 [9] suggests that the base band centre frequency shall use a Fourier index q multiple of 32
when mapped to the DFT indexes. Specifically is recommended to:
a) assign the middle carrier to the half-way index q = N/2, i.e. the half-sampling-frequency term; or
b) assign the middle carrier to index q = 0, i.e. the DC or zero-frequency term.
As both alternatives produce the same result, alternative b is used here for calculate what happens to the outermost
continual pilots in each DVB-T mode when adding the corresponding Guard Intervals.
For the useful part of the OFDM symbol all carriers are orthogonal, hence all have an integer number of cycles. When
the guard interval is added, the orthogonality does not apply to the total length of the symbol.
NOTE: The orthogonality is regained in the demodulator when the appropriate time window is selected for
demodulation.
The value of the indexes for the outermost continual pilots are: q = -3 408 and q = +3 408 for the 8k mode being
q = –852 and q = +852 for the 2k mode. The number of cycles per GI is tabulated below.
Table E.1
8k mode (Pilots k = 0 and k = 6 817) 2k mode (Pilots k = 0 and k = 1 704)
Cycles · Guard
Interval
3 408 × 1/4 3 408 × 1/8 3 408 ×1/16 3 408 × 1/32 852 ×1/4 852 ×1/8 852 × 1/16 852 × 1/32
Number of
cycles
852 426 213 106,5 213 106,5 53,25 26,625
The continual pilots are modulated according to a PRBS sequence, wk, corresponding to their respective carrier index k.
The PRBS is initialized so that the first output bit from the PRBS coincides with the first active carrier. This means that
the PRBS is initialized at each new symbol and then each continual pilot has assigned in each symbol the same phase as
it had in the precedent symbol, so for the continual pilots that have an integer number of cycles in the guard interval, it
would not be any phase change from one symbol to the next.
This happens as per the above table to the two outermost continual pilots when the guard intervals of ¼, 1/8 or 1/16 are
used in the 8k mode or when the ¼ is used in the 2k mode. That is why only in these cases the outermost continual
pilots are represented as single spectral lines in the spectrum analysers.
Note that the central carrier is always multiple of 32 as per the specified recommendation, however the central carrier is
a continual pilot only in the 8k mode, while in the 2k mode it is a data carrier that changes phase according to the data
being transmitted in each symbol. The central carrier in 8k mode is always seen as a single spectral line on a
spectrum analyser.
Figure E-2: Examples of 8k centre channel measurement with sweeping spectrum analyser Ch 68,
and with digital spectrum analyser CH 69
NOTE: Centre of screen was selected at the nominal pilot position.
ETSI
100 ETSI TR 101 290 V1.2.1 (2001-05)
E.1.2 Measurement in other cases
When the outermost carriers or the central carrier can not be conveniently used for frequency measurements, it is
possible to find a continuous pilot carrier that shows a single spectral line in the spectrum, so it can be measured with
the counter of the spectrum analysers.
The continual pilot k = 48 does have the property for the 8k mode to have integer number of cycles in all GI, but not for
the 2k mode.
q = -3 408 + 48 = -3 360
The continual pilot k = 1 140 is the only one for the 2k mode that has the property of having an integer number of
cycles in all GI
q = -852 + 1 140 =288
See table E.2.
Table E.2
8k mode (Pilot k = 48) 2k mode (Pilot k =1 140)
Cycles × Guard
Interval
3 360 × 1/4 3 360 × 1/8 3 360 × 1/16 3 360 × 1/32 288 × 1/4 288 × 1/8 288 × 1/16 288 ×1/32
Number of
cycles
840 420 210 105 72 36 18 9
The following formulas can be used for calculate the central frequency of the channel Fc:
In 8k mode: Fc = Fkmeasured + [(3 408 – k) × Fspacing]
In 2k mode: Fc = Fkmeasured + [(852 – k) × Fspacing]
The following examples are valid for 8 MHz Channels. Similar example calculations may be made for 7 MHz and
6 MHz channels.
Figure E-3: Examples of 8k carrier k = 48 and 2k carrier k = 1 140 measurements on CH 69
NOTE: Centre of screen was selected at the nominal pilot position.
The continual pilot k = 48 in 8k mode also has the property that is located at exactly -3,75 MHz from the centre of the
channel, making it very convenient for the measurement.
The carrier k = 1 140 for the 2k mode however does not fall at any easy-to-remember frequency; it has a frequency
offset of +1,285 714 28 MHz.
ETSI
101 ETSI TR 101 290 V1.2.1 (2001-05)
If channel 69, for example, is modulated in 8k mode and its carrier k = 48 is being measured as 854,250 015 63 MHz,
then the centre of the channel is: Fc = 854 250 015,63 + [(3 408 – 48)· 1 116,0715] = 858 000 015,63 Hz.
If channel 69, for example, is modulated in 2k mode and its carrier k=1140 is being measured as 859,285 729 63 MHz,
then the centre of the channel is: Fc = 859 285 729,63 + [(852 – 1 140) × 4 464,2857] = 858 000 015,35 MHz.
In case of 2k mode when using guard intervals greater than 1/32, a suitable carrier for centre channel measurements is
the k = 804, which happens to lie at CF - 1MHz + 785 714 Hz. This is easy to measure and calculate, and is closer than
carrier 1 140 to the centre of channel (less than 250 kHz).
E.1.3 Calculation of the external pilots frequency when they do
not have continual phase.
It is worth to remember that in the DVB-T modulation mode and due to the insertion of the guard interval, the
frequency spacing does not equal the width of the lobes of modulated carriers.
The frequency spacing is founded as the inverse of the useful part interval of the mode used. For example in a 8 MHz
channel system, the 2k mode has useful interval of TU = 224 μs, thus the frequency spacing is:
- Fs = 1/224 μs = 4 464,285714...Hz and for the 8k mode the corresponding values are: TU = 896 μs; and
- Fs = 1/896 μs = 1 116,071 429...Hz. (similar calculations are valid for channel bandwidths other than 8 MHz).
The width of the side-lobes is found as the inverse of the total symbol length of the mode and guard interval used, the
main lobe has twice the width of the side lobes. Four cases are found for measurements.
Table E-3 indicates the corresponding values.
Table E.3
8, 7 and 6 MHz Channels 8k mode 2k mode
Guard Interval 1/4 1/8 1/16 1/32 1/4 1/8 1/16 1/32
(8MHz) TS = Δ + TU (μs) 1 120 1 008 952 924 280 252 238 231
Side lobe width 1/TS (Hz) 892,8571 992,0635 1 050,4202 1 082,2511 3 571,4286 3 968,2540 4 201,6807 4 329,0043
(7 MHz) TS = Δ + TU (μs) 1 280 1 152 1 088 1 056 320 288 272 264
Side lobe width
1/TS (Hz)
781.25 868,0556 919,1176 946,9697 3 125 3 472,2222 3 676,4706 3 787,8787
(6 MHz) TS = Δ + TU (μs) 1 493,3 1 344 1 269,3 1 232 373,3 336 317,3 308
Side lobe width 1/TS (Hz) 669,65 744,04 787,83 811,68 2 678,81 2 976,19 3 151,59 3 246,75
Number of cycles 852 426 213 106.5 213 106,5 53,25 26,625
Measurement Case A A A B A B C D
ETSI
102 ETSI TR 101 290 V1.2.1 (2001-05)
Measurement case A: for the 8k mode at ¼, 1/8 and 1/16 as well as for the 2k mode at ¼, as there are single spectral
lines, the outermost pilots are orthogonal for the symbol length as has been seen above, the pilot frequency is measured
directly in these cases. For example Fp = 861 803 586 Hz, or Fp = 861 803 617 Hz as indicated below for a channel 69
measurement on system G of 8 MHz.
Figure E-4: Examples of 2k carrier k = 1 704 and 8k carrier k = 6 816 measurements, both at ¼ Guard
Interval on CH 69 (different day and different error)
NOTE 1: Centre of screen was selected at the nominal pilot position.
In the other cases, and due to the non-orthogonality of the pilots for the total symbol length, the pilots shown a Fourier
series of lines whose amplitude and frequency depends on the phase and size of the truncation of the pilot in the period
of the symbol. These frequencies are equally spaced at the inverse of the lobe width (total symbol length).
Measurement case B: the cases of 8k mode at 1/32 and 2k mode at 1/8 shows that the truncation of the sinusoidal
cycles is 0,5 cycles. This means that two symmetrical spectral lines can be found around the central position (expected
pilot position). The central position can be found as the mean of the two frequencies when they are measured.
Another calculation mode for this case is to measure one of the two spectral lines and add or subtracts half of lobe width
(1/symbol-length).
For example, in 8 MHz system, if the lower one of the two lines is measured as Fh = 861 803 083,50 Hz, then the
calculated frequency of the corresponding external pilot would be Fp = 861 803 083,50 + 1,082.25/2 = 861 803 624,6
Hz for the 8k mode, or similar calculation may be done for the 2k mode, also as example, in 8 MHz channel system,
Fp = 861 801 602,25 + 3 968,25/2 = 861 803 586,38 Hz.
(Measured values are in italics, nominal values are in normal text).
Figure E-5: Examples of 8k carrier k = 6 816 at 1/32 GI and 2k carrier k = 1 704 at 1/8 GI
measurements, on CH 69 (different day and different error)
ETSI
103 ETSI TR 101 290 V1.2.1 (2001-05)
NOTE 2: Centre of screen was selected at the nominal pilot position.
Measurement case C: the case for the 2k mode at 1/16 is a bit more complex, the truncation happens at 0,25 cycles. In
this case the highest amplitude spectral line is located at ¼ the lobe width above the nominal position of the pilot (for
the lower pilot) or at ¼ the lobe width below (for the upper pilot).
If this highest amplitude line frequency is measured as Fh = 854 197 491 Hz, the lower pilot frequency is calculated as
Fp = 854 197 491 - 4 201,68/4 = 854 196 440 Hz.
If this highest amplitude line frequency is measured as Fh = 861 802 539 Hz, the upper pilot frequency is calculated as
Fp = 861 802 539 + 4 201,68/4 = 861 803 590 Hz.
Figure E-6: Examples of 2k carrier k = 0 and carrier k = 1 704 at 1/16 GI measurements, on CH 69
NOTE 3: Centre of screen was selected at the nominal pilot position.
As per definitions in 9.1.2 RF channel width (Sampling frequency accuracy) the following results are found:
• The RF channel width for this channel 69 of system G (8 MHz) is calculated as:
- 861 803 590,5 - 854 196 440,7 = 7 607 149,8 Hz, that is 7 Hz wider than nominal.
• The sampling frequency of the modulator is calculated as:
- 7 607 149,8 × 4 096/1 704 = 18 285 730,9 Hz, that is 16,6 Hz higher than expected. Or it may be said that
the accuracy is: 16,6/18 285 714,28 = 9,13 × 10-7 or 0,913 ppm.
Measurement case D: The case for the 2kmode at 1/32 is also somewhat complex, the truncation happens at
0,625 cycles. For the lower pilot, one spectral line falls at 5/8 the lobe width above the nominal position of the pilot and
the other line, the highest in amplitude falls at 3/8 the lobe width below the nominal position of the pilot. That is at
62,5 % above and 37,5 % below respectively. For the upper pilot the highest line falls 3/8 above nominal position and
the other line falls at 5/8 below nominal.
If the highest-level (lower in frequency) signal is measured for the lower pilot as Fh = 854 194 819 Hz then the pilot
frequency is calculated as Fp = 854 194 819 + 4 329 × 3/8 = 854 196 442 Hz.
ETSI
104 ETSI TR 101 290 V1.2.1 (2001-05)
If the highest-level (upper in frequency) signal is measured for the upper pilot as Fh = 861 805 211 Hz then the pilot
frequency is calculated as Fp = 861 805 211 - 4 329 × 3/8 = 861 803 588 Hz.
Figure E-7: Examples of 2k carrier k = 0 and carrier k = 1 704 at 1/32 GI measurements, on CH 69
NOTE 4: Centre of screen was selected at the nominal pilot position.
As per definitions in 9.1.2 RF channel width (Sampling frequency accuracy) the following results are found:
• The RF channel width for this channel 69 of system G (8MHz) is calculated as:
- 861 803 588,6 - 854 196 442,6 = 7 607 146 Hz, that is 3,1 Hz wider than nominal.
• The sampling frequency is calculated as:
- 7 607 146 × 4 096/1 704 = 18 285 721,84 Hz, that is 7,56 Hz higher than expected. Or it may be said that the
accuracy is: 7,56/18 285 714,28 = 4,134 × 10-7 or 0,413 ppm.
ETSI
105 ETSI TR 101 290 V1.2.1 (2001-05)
The offset values for all four measurement cases are summarized in table E.4.
Table E.4
8, 7 and 6 MHz Channels 8k mode 2k mode
Guard Interval 1/4 1/8 1/16 1/32 1/4 1/8 1/16 1/32
(8 MHz) TS = Δ + TU (μs) 1 120 1 008 952 924 280 252 238 231
Side lobe width 1/TS (Hz) 892,8571 992,0635 1 050,4202 1 082,2511 3 571,4286 3 968,2540 4 201,6807 4 329,0043
Add or subtract Hz 0 0 0 ±541 Hz 0 ±1 984 Hz ±1 050 Hz ±1 623 Hz
(7 MHz) TS = Δ + TU (μs) 1 280 1 152 1 088 1 056 320 288 272 264
Side lobe width 1/TS (Hz) 781,25 868,0556 919,1176 946,9697 3 125 3 472,2222 3 676,4706 3 787,8787
Add or subtract Hz 0 0 0 ±473 0 ±1736 ±919 ±1 420
(6 MHz) TS = Δ + TU (μs) 1 493,3 1 344 1 269,3 1 232 373,3 336 317,3 308
Side lobe width 1/TS (Hz) 669,65 744,04 787,83 811,68 2 678,81 2 976,19 3 151,59 3 246,75
Add or subtract Hz 0 0 0 ±406 0 ±1488 ±788 ±1 218
NOTE 5: The values for 2k with GI of 1/16 are to be added or subtracted to the highest of the two spectral lines
around the nominal position of the upper or lower pilot respectively (1/4 factor), the values for 2k with GI
of 1/32 are to be added or subtracted to the highest of the two spectral lines around the nominal position
of the lower or upper pilot respectively (3/8 factor).
E.1.4 Measuring the symbol length and verifying the Guard
Interval
If appropriate span and average is used when analysing the spectrum of a DVB-T signal, it is possible to display the
scattered pilots to a detail that may be used to measure the interval between 4 OFDM symbols.
NOTE: The definition for the elementary interval provides the useful duration of the symbol as:
- For 2k the useful interval is TU = 2 048 × Ep;
- For 8k the useful interval is TU = 8 192 × Ep.
See table E.5.
Table E.5
8 MHz 7 MHz 6 MHz
8k 2k 8k 2k 8k 2k
Elementary period: Ep 7/64 (μs) = 0,109 375 μs 8/64(μs) = 0,125 μs 7× (4/3)/64 (μs) = 0,145 833 3…μs
Useful duration: TU 896 μs 224 μs 1024 μs 256 μs 1 194,6666…μs 298,6666…μs
ETSI
106 ETSI TR 101 290 V1.2.1 (2001-05)
In figure E-8 seven data carriers (k = 6 809 through k = 6 815), two scattered pilots (k = 6 810 and k = 6 813) and the
upper pilot (k = 6 816) are seen at a 10 kHz total span for an 8 MHz channel. The effect of the scattered pilots can be
easily seen every three carriers in the frequency axis. Each scattered pilot has always the same phase for a given
location, then it behaves as a burst of a fixed frequency and phase that repeats every four OFDM symbols and has
duration of one symbol. The spectra created by the scattered pilots overlaps with the spectra of the data carrier
associated in the same location, which appears over three consecutive symbols between the appearances of the scattered
pilot itself. The spectrum of the data carriers is a lobular dense spectrum due to the QAM modulation that changes from
symbol to symbol.
Figure E-8: Examples of 8k carrier k = 6 813 and at 1/4 GI measurements, on CH 69
NOTE: Centre of screen was selected at the nominal pilot position.
Due to the characteristics explained above, the scattered pilots present a line spectrum with lobular envelope. For this
kind of sinusoidal pulsed signal with fixed phase and frequency at the start of each RF pulse, the width of the lobes is
the inverse of the duration of a symbol (i.e. 1/1 120 = 892,85 Hz for a 8 MHz channel, 8k and ¼ GI as indicated on
table E.6). However this lobe width is not easily measurable. The separation of the spectral lines is the inverse of the
repetition period of the scattered pilot occurrence (i.e. ¼ 480 = 223,2 Hz for same example as before). These lines can
easily be measured with currently available instruments. Detailed measurement at 500 Hz total span, shows that even
one of the most demanding cases, the 8 k mode at ¼ GI with line separation of 223,2 Hz can be measured as indicated
at right.
The line separation that can be expected for the different DVB-T modes, is detailed in tables E.6, E.7 and E.8.
Table E.6
8 MHz Channels 8k mode 2k mode
Guard Interval 1/4 1/8 1/16 1/32 1/4 1/8 1/16 1/32
TS = Δ + TU (μs) 1 120 1 008 952 924 280 252 238 231
Scattered repetition period μs 4 480 4 032 3 808 3 696 1 120 1 008 952 924
Line spectra separation Hz 223,2 248 262,6 270,6 892,9 992,1 1 050,4 1 082,3
ETSI
107 ETSI TR 101 290 V1.2.1 (2001-05)
Table E.7
7 MHz Channels 8k mode 2k mode
Guard Interval 1/4 1/8 1/16 1/32 1/4 1/8 1/16 1/32
TS = Δ + TU (μs) 1 280 1 152 1 088 1 056 320 288 272 264
Scattered repetition
period μs
5 120 4 608 4 352 4 224 1 280 1 152 1 088 1 056
Line spectra separation Hz 195,3 217 229,8 236,7 781,3 868,1 919,1 947
Table E.8
6 MHz Channels 8k mode 2k mode
Guard Interval 1/4 1/8 1/16 1/32 1/4 1/8 1/16 1/32
TS = Δ + TU (μs) 1 493,3 1 344 1 269,3 1 232 373,3 336 317,3 308
Scattered repetition
period μs
5 973,3 5 376 5 077,3 4 928 1 493 1 344 1 269 1 232
Line spectra separation
Hz
167,4 186 197 202,9 669,6 744 787,8 811,7
Measuring the line spacing of the scattered pilots and checking against the above table will provide the answer to which
is the actual Guard Interval and mode being embedded in the measured spectrum.
Notice that: for the cases where the outermost continual pilots do not have continuous phase as indicated above in E.1.3,
the distance between two spectral lines can be checked against table E.3 to verify what symbol length is being used, and
consequently what Guard Interval is being used.
Figure E-9 has two measurement examples, one where the span has been set to 10 kHz and the separation of two
spectral lines of a scattered pilot is 890,63 Hz as indicated by the delta marker. The nearest figure in table E.6 is
892,9 Hz thus it can be inferred this case is a 2 k mode at ¼ GI. The figure at right, with span at 2 kHz, shows a line
separation of 1084,38 Hz, corresponding to 2k mode at 1/32 GI (1 082,3 in table E.6).
Figure E-9: Examples of 2k carrier k = 1 701 at ¼ and at 1/32 GI measurements, on CH 69
NOTE: Centre of screen was selected at the nominal pilot position.
ETSI
108 ETSI TR 101 290 V1.2.1 (2001-05)
E.1.5 Measuring the occupied bandwidth, and calculation of the
frequency spacing and sampling frequency
The occupied bandwidth depends directly from the frequency spacing and this from the sampling frequency.
If the frequency of the external pilots is known, see above on how to measure them, then the related values may be
calculated as per table below. Denoting the outermost pilot frequencies as FL and FH appropriately the occupied
bandwidth is OB = FH _ FL. The number of carriers K, and for 2k mode K-1 = 1 704 while for 8k mode K-1 = 6 816.
Table E.9
Calculated value Nominal value (8 MHz Channels)
8k mode 2k mode 8k mode 2k mode
Occupied bandwidth FH - FL 7,60714285714285714285714285714286… MHz
Frequency Spacing (FH - FL)/6 816 (FH - FL)/1 704 1 116,0714285…Hz 4 464,2857142…Hz
Useful duration 6 816/(FH - FL) 1 704/(FH - FL) 896 μs 224 μs
Centre channel 1st IF (FH - FL) × 4 096/(K-1) (FH - FL) × 1 024/(K-1) 4,57142857142857142857142857142857…MHz
Sampling Frequency (FH - FL) × 16 384/(K-1) (FH - FL) × 4 096/(K-1) 18,2857142857142857142857142857143…MHz
NOTE: The long periodic decimals have been calculated using the Calculator facility from Windows, and have
been left here as resulted from copying through the clipboard, as a matter of curiosity only.
Values in italics are approximate values.
Table E.10
Nominal value (7 MHz Channels) Nominal value (6 MHz Channels)
8k mode 2k mode 8k mode 2k mode
Occupied bandwidth 6.656250 MHz 5,70535714285714285714285714285842… MHz
Frequency Spacing 976,5625 Hz 3 906,25 Hz 837,053571428571…Hz 3 348,2142857142…Hz
Useful duration 1 024 μs 256 μs 1 194,666666…μs 298,666666…μs
Centre channel 1st IF 4 MHz 3,42857142857142857142857142857334…MHz
Sampling Frequency 16 MHz 13,7142857142857142857142857142934…MHz
E.2 Selectivity
See clause 9.2.
DVB-T
Test transmitter
CW
Signal Generator
DVB-T
Rx
BER
Monitor
N
W, X
DUT
Figure E-10: Selectivity
E.3 AFC capture range
See clause 9.3.
ETSI
109 ETSI TR 101 290 V1.2.1 (2001-05)
D V B -T
R x
M PEG -2
T S A n a lys e r
D U T
D V B -T N Z
T e s t tra n sm itte r
Figure E-11: AFC capture range
E.4 Phase noise of Local Oscillators (LO)
See clause 9.4.
The measurement can be done with a spectrum analyser. As the spectrum shape of the phase noise sidebands of any
Local Oscillator (LO) used in the process of up/down conversion could be very different depending on factors such as
the type of crystal cut, the filter of the PLL, the noise of the active devices involved, etc. it is not convenient to integrate
the spectrum of a sideband to reflect a single measured number which could not have meaning at all.
However, samples at certain offsets of the oscillator signal could have more meaning, as indicated in clause 9.4. In each
case of Common Phase Error (CPE) and Inter-Carrier Interference (ICI), 3 frequencies at each side of the oscillator
signal should be measured. In order to make the measurement as accurate in frequency as possible, the spectrum
analyser should be set to the minimum resolution filter available, and should be, at least, as low as 1 kHz for the 2 k
system and 300 Hz for the 8 k system. In order to average the noise, the video filter should be activated with a value of
at least 100 times narrower than the resolution filter used. The measured values should be normalized to a 1 Hz
bandwidth.
Should the spectrum analyser used not have the 1 Hz normalization capability, it can be done manually with the
following criterion:
For example: carrier frequency: 36 MHz
fm = 10 kHz (represents any of the required offsets fa, fb or fc)
ΔB = Equivalent Noise Bandwidth (ENB) of the resolution bandwidth filter: 270 Hz
video bandwidth: 10 Hz or 30 Hz
NOTE 1: The spectrum analysers typically use near Gaussian filters for the resolution bandwidth with a 20 %
tolerance. The Equivalent Noise Bandwidth (ENB) is equal to the bandwidth of the filter measured at
-3,4 dB, (by actually measuring the filter of the spectrum analyser, the 20 % tolerance factor is
eliminated).
Then the following conversion to 1 Hz bandwidth can be applied:
Pn ≅ (noise _ power _ in _ ΔB)dBm−10log10ΔB + 2,5dB in [dBm/Hz]
NOTE 2: The 2,5 dB term accounts for the correction of 1,05 dB due to narrowband envelope detection and the
1,45 dB due to the logarithmic amplifier.
E.4.1 Practical information on phase noise measurements
This example from the works of AC106 VALIDATE Project and taken from the DTG D book, shows a recommended
mask for phase noise measurements that is valid for local oscillators and is considered to cover safe limits for both CPE
and ICI phase errors in the 2k mode of DVB-T. The following values are recommended.
ETSI
110 ETSI TR 101 290 V1.2.1 (2001-05)
Table E.11: Frequency offsets for phase noise measurements
fa fb fc fd
Frequency 10 Hz 100 Hz 3 kHz 1 MHz
Limits La to Ld -55 dBc/Hz -85 dBc/Hz -85 dBc/Hz -130 dBc/Hz
Figure E-12: Example for phase noise mask
The total phase noise in the signal is the cumulative effect of all local oscillators (L.O.) that are used in the signal path.
Clause A.4 can be seen for additional information on phase noise measurements.
E.5 RF/IF signal power
See clause 9.5.
The signal power can be measured directly at the interfaces K, L, M, N or P or by using a calibrated splitter. Care
should be taken at interfaces L or Mnot to overdrive the maximum allowed input signal for the spectrum analyser or
power metre.
The shoulders of the spectrum should not be accounted for in the measurement of power because them do not represent
any useful power conveying information. The shoulders are unwanted results of the FFT process and also due mainly to
non-linearity of the practical implementations.
E.5.1 Procedure 1 (power metre)
An spectrum analyser is used with an integrating routine which can measure the mean power along frequency slots
covering the overall part of the spectrum to be measured (this capability is currently available in several spectrum
analyser on the market). In this case the values to be supplied to such a routine or to be used if manual undertaken of the
measurement is wanted are:
1) Centre frequency of the spectrum: if possible as calculated under measurement E.2;
2) Spectrum bandwidth of the signal: 7,61 MHz for an 8 MHz channel system.
A
B C
fa fb fc
-La
-Lb & Lc
-Ld
0 dB
0 Hz
Frequency offsetts
Carrier
1.- Possible mask for phase noise measurements. The axis are not to scale, see table for values.
D
fd
ETSI
111 ETSI TR 101 290 V1.2.1 (2001-05)
E.5.2 Procedure 2 (spectrum analyser)
With the above considerations in mind, it would be very difficult to use an exact square filter for the measurement with
a power sensor, however a good approximation should be obtained if a filter is used which can even take in account part
of the shoulders in the measurement.
For measuring with a thermal power sensor such an appropriate filter should be used.
Figure E-13: Test set-up for RF/IF power measurement
E.6 Noise power
See clause 9.6.
Typically all the power present in a channel which is not part of the signal can be regarded as unwanted noise. It can be
produced from different origination and be of the form of random noise (thermal), pseudo-random (digitally modulated
interfering carriers) or periodic (Continuous Waves CW or narrowband interference), the first two are called
non-coherent and the periodic ones are termed as coherent. In this measurement, all different types of noise are
measured simultaneously, and the measured result can be termed as unwanted power.
For doing this measurement the signal shall be switched off. The measurements can be done at interface N (RF level) or
at interface P (IF level).
Noise level can be measured with a spectrum analyser or any other appropriate device. The same bandwidth
considerations and methodology used in clause E.6 apply to this measurement in both cases, using a power metre and a
spectrum analyser.
Figure E-14: Test set-up for out-of-service noise power measurement
E.6.1 Procedure 1
Exactly equal to the above preferred procedure for signal power, clause E.6, but understanding that the signal for this
channel under measurement has been switched off.
E.6.2 Procedure 2
Using a power metre as in the alternate procedure above in clause E.6, using the same filter and with the channel signal
off.
E.6.3 Procedure 3
If the noise floor in all bandwidth of interest is flat, it would be possible to measure the noise power at any frequency
point inside the channel bandwidth and normalize the value to the nominal bandwidth of (n-1) × fSPACING (7,61 MHz
for 8 MHz channels 6,66 MHz for 7 MHz channels).
ETSI
112 ETSI TR 101 290 V1.2.1 (2001-05)
If the spectrum analyser does not have normalization routine to the wanted bandwidth the following procedure can be
used.
In order to average the noise, the video filter should be activated with a value of at least 100 times narrower than the
resolution filter used, this resolution bandwidth filter should be chosen to be as wide as possible in order to average as
much spectrum of the channel as possible, but not exceeding such bandwidth (e.g. 7,61 MHz), the equivalent noise
bandwidth ΔB (MHz) of the filter should be known by the specifications given by the manufacturer, or measured
following manufacturer indications. The noise power measured can be normalized to the wanted bandwidth using the
following formulae:
Noise power (dB) = Measured level (dB) + 10 log10 (7,61/ΔB) + 2,5 dB (for 8 MHz channels)
If the spectrum analyser has a routine to normalize to 1 Hz, (this use to include the 2,5 dB correction) but not able to
normalize to the wanted bandwidth, the following conversion can be applied:
Noise power (dB) = Measured level (dB/Hz) + 10 log10 (7,61 × 106) =
Measured level (dB/Hz) + 68,8 dB (for 8 MHz channels)
E.6.4 Measurement of noise with a spectrum analyser
Care should be taken when the measured noise has a display level close to the display level of instrument noise, (less
than 10 dB), because an additional proximity factor should be applied. This is typically done automatically in some
instruments available in the market.
If this is not available in the instrument, it is necessary to subtract a correction factor CF from the noise level measured,
the following correction table can be used.
Table E.12: Correction Factor (CF) for measured noise level
D (dB) CF (dB)
0,5 8,63
1 6,87
1,5 5,35
2 4,33
3,01 3,01
4 2,2
5 1,65
6 1,26
7 0,98
8 0,75
9 0,58
10 0,46
D is the distance in display level between the instrument noise (no signal applied to the input) and measured noise level
(with no change in the settings).
Notice that below 2 dB of D, the reliability of the result after applying the CF is under question due to the uncertainty of
the measurement and the corresponding big value of CF to be subtracted.
E.7 RF and IF spectrum
See clause 9.7.
To be defined after some practical experience is achieved.
ETSI
113 ETSI TR 101 290 V1.2.1 (2001-05)
E.8 Receiver sensitivity/dynamic range for a Gaussian
channel
See clause 9.8.
N
W, X
DUT
DVB-T
Test transmitter
DVB-T
Rx
BER
Monitor
Figure E-15: Receiver sensitivity/dynamic range for a Gaussian channel
E.9 Equivalent Noise Degradation (END)
See clause 9.9.
N, P, S
W, X
DVB-T
Tx
DVB-T
Rx
No ise
Generator
BER
Monitor
Figure E-16: Equivalent Noise Degradation (END)
All measurements of performance parameters are carried out by using a dummy load which provides a return loss for
the wanted channel which is low enough not to influence the measurement.
E.9.1 Description of the measurement method for END
To improve the accuracy of the measurement, two independent noise sources are used. By this, the influence of the
tolerance of the first attenuator is eliminated which could well be in the same magnitude as the wanted measurement
result.
The following steps should be carried out to arrive at an accurate ENF value:
1) Connect the real DVB-T transmitter to the DVB-T receiver and add Gaussian noise, Ncal, to the point where the
BER reaches a pre-determined value (e.g. 2 x 10-4 after Viterbi decoding). Ncal does not have to be measured.
No channel noise, Nch, should be added. The C/N at the input to the receiver (Interface C) is therefore
C/(Ntx + Ncal).
2) Replace the real DVB-T transmitter by the ideal one (disconnect Ntx in figure E-17). The C/N at Interface C is
now somewhat higher (C/Ncal), since Ntx is no longer present. The BER is therefore now lower than the
predetermined value.
ETSI
114 ETSI TR 101 290 V1.2.1 (2001-05)
3) Add Gaussian channel noise, Nch, to the point where the BER has reached its predetermined value again. The
C/N at interface C is now C/(Nch + Ncal).
4) Measure the value of C/Nch at Interface B.
+ + +
Ideal
DVB-T
Tx
Ncal Ntx Nch
Unknown
DVB-T
Rx
Real DVB-T Tx
Fixed
BER=2 x 10-4
after Viterbi or
RS decoding
C
Interface
A
Interface
B
Interface
C
Figure E-17: ENF measurement scheme
Since both C/(Ntx + Ncal) and C/(Nch + Ncal) lead to the same BER, Nch can be identified with Ntx and be regarded as
an estimate of Ntx.
The ENF is defined to be 10 10log(Ntx/C). The estimated ENF value is similarly 10 10log(Nch/C)
As long as all distortions of a DVB-T transmitter can be well approximated by the Gaussian noise, Ntx, the ENF
measurement, as described above, should be completely independent of both the DVB-T mode and the receiver
characteristics. For highest measurement accuracy the measurement should however preferably be done using the
(non-hierarchical) mode requiring the highest C/N, i.e. 64-QAM R=7/8.
In practice, there might however be selective effects such as amplitude ripple and spurious signals within the useful
bandwidth. In these cases the ENF will in principle be better (= a more negative value) when stronger code rates are
used (such as R = 1/2 or 2/3) than when weaker codes are used (such as R = 5/6 or 7/8). Whether this difference is
measureable or not remains to be seen. It is therefore recommendable to measure the ENF also for the other code rates.
If there is negligeable difference between the ENF figures for the different code rates, this will imply that there are few
selective effects and/or that these effects can be well approximated by Gaussian noise. If however there is a significant
difference in ENF figures this implies that the ENF (and hence END) is code rate dependent. In such a case the ENF
value to be used (either by itself or for the calculated END) should preferably be the one measured with the same code
rate as the DVB-T transmitter will be used with by the network operator.
E.9.2 Conversion method between ENF and END
Let (C/N)min, theory be the minimum C/N requirement for a DVB-T mode given by EN 300 744 [9].
Assume an implementation loss of 3,0 dB for all modes.
Let X = (C/N)min, real be the corresponding minimum required C/N for a DVB-T mode.
X = (C/N)min, real = (C/N)min, theory + 3,0 dB
END can be calculated from ENF by the formula:
END = -10 10log(10 -X/10 -10 ENF/10) - X
Example:
ETSI
115 ETSI TR 101 290 V1.2.1 (2001-05)
X = 19,5 dB (64QAM, R= 2/3) ENF = -30,0 dB
END = -10 10log(10 -19,5/10 -10 -30,0/10) - 19,5 dB = 0,41 dB
E.10 Linearity characterization (shoulder attenuation)
Figure E-18: Test set-up for "linearity characterization"
E.10.1 Equipment
(1) OFDM signal source (interface K or L of DVB-T transmitter);
(2) attenuator, possibly adjustable in 0,1 dB (max. 0,5 dB) steps. Optional, see clause E.10.2, remark (d);
(3) transmitter under measurement;
(4) power attenuator;
(5) directive coupler or attenuator, see clause E.10.2, remark (a);
(6) spectrum analyser;
(7) attenuator, possibly adjustable. Optional, see clause E.10.2, remark (c);
(8) power metre. Optional, see clause E.10.2, remark (a).
E.10.2 Remarks and precautions
(a) Power metre (8) can be useful to verify and monitoring the output power of the transmitter (3) and for the
calibration process. If power metre (8) is not available, the directive coupler (5) can be replaced by an opportune
attenuator connected to the spectrum analyser (6).
(b) Care should be taken in the choice of the power attenuator (4) in terms of max. admitted power.
(c) Care should be taken in the choice of all attenuators (and directive coupler) to prevent damage to test-set
equipment. For example, the function of the optional attenuator (7) is to protect the probe of the power metre.
ETSI
116 ETSI TR 101 290 V1.2.1 (2001-05)
The attenuator (7) can also be useful for other measurements and, for example, be connected in a chain to the
receiver.
(d) Pay attention to the admitted power at the IF (or RF) input of the transmitter, in order to obtain a proper working
point. Optional attenuator (2) can be used for this purpose.
E.10.3 Measurement procedure (example for UHF channel 47)
- Step 1: Select the centre frequency of spectrum analyser in the middle of the UHF channel (i.e. 682 MHz for
channel 47). Verify the output power level using an high resolution BW (3 MHz or 5 MHz) and
compare with the value obtained by the power metre (if available).
- Step 2: Select the centre frequency of spectrum analyser at the end of the UHF channel (i.e. 686 MHz for
channel 47).
- Step 3: Select an adequate span (for example 2 MHz).
- Step 4: Select the resolution BW (10 kHz is adequate for 2 k and 8 k mode) and adjust levels.
Video BW is of the same order.
- Step 5: Measure the power level at 300 kHz and 700 kHz from upper edge of the DVB-T spectrum and
proceed as indicated in clause 9.10. Last DVB-T carrier is at approximately +3,8 MHz from the centre
of the UHF channel: then, for channel 47, the two measurement points are at 686,1 MHz and
686,5 MHz.
- Step 6: Repeat steps from 2 to 5 for the lower edge of the spectrum.
- Step 7: The worst case value of the upper and lower results is the "shoulder attenuation" (dB).
NOTE: The value obtained should be joined up with the used mode (2 k or 8 k) of the OFDM source.
If available, the "maximum-hold" function of the spectrum analyser can help to carry out the measurement.
685 685,5 686 686,5 687
Power
[dBm]
Res. BW10 kHz
VideoBW10 kHz
Span 2MHz
Shoulder
attenuation
+300 kHz +500kHz +700 kHz
Endof UHFchannel 47
LastDVB-Tcarrier
(at approx. 685,8MHz)
DVB-T spectrum
(max. value)
ref. 0 kHz
DVB-Tspectrum
Frequency
[MHz]
Figure E-19: Example with the upper edge of the DVB-T spectrum in UHF channel 47
ETSI
117 ETSI TR 101 290 V1.2.1 (2001-05)
E.11 Power efficiency
DVB-T
Tx
Mains
Power
Meter
RF Power
Meter
A M
Figure E-20: Power efficiency
E.12 Coherent interferer
Connect a suitable spectrum analyser to interface N.
E.13 BER vs. C/N by variation of transmitter power
DVB-T
Tx
Noise
Generator
RF Power
Meter
M
PRBS
Generator
E, F
BER
Test Set
DVB-T
Test Receiver
N, P, R U, V
N
Figure E-21: BER vs. C/N by variation of transmitter power
Adjust signal level at receiver input to the same value for different Tx output power values by attenuator.
The results of this measurement can be put in diagrams, such as:
- BER vs. C/N for constant Pout;
- BERvs.Pout for constant C/N;
- BERvs.Pout for constant noise power.
ETSI
118 ETSI TR 101 290 V1.2.1 (2001-05)
E.14 BER vs. C/N by variation of Gaussian noise power
Figure E-22: BER vs. C/N by variation of Gaussian noise power
E.15 BER before Viterbi (inner) decoder
See clause 9.15.
NOTE: For the measurements described in clauses 9.15, 9.16, 9.17, 9.18 and 9.19 dedicated measurement
instruments are envisaged.
E.16 Overall signal delay
The set-up for measurement delay of transmitters by using a reference transmitter is illustrated in figure E-23, on which
the adjustable delay in the reference transmitter, is optional.
It is intended that the reference transmitter be built with as minimum delay as possible. With this in mind there are two
possible ways of measuring the difference of delay between the transmitter under test and the reference transmitter.
a) Directly from the measurement of the width of the lobes as illustrated in figure E-24. The estimated delay
measured graphically in this figure is 770 ns.
b) By inserting a calibrated variable delay in the reference transmitter as illustrated in figure E-23. The delay is
then increased by steps until the width of the lobes is high enough to be greater than the width of the channel.
Then the difference in delay is that of the inserted one (figure E-25).
NOTE: When the lobe width is exactly 8 MHz, the relative delay is 1/8 = 125 ns. If wider lobe is achieved, less
relative delay is present. These range of delays represent a minimal fraction of the guard interval and
consequently no higher accuracy is typically needed.
The shortest guard interval for 8 MHz channel corresponds to 7 μs (1/32 of 224 μs) in the 2k mode.
Figure E-25 shows a case where the delay was adjusted until the width of the lobe was greater than the channel
width, being the delay less than 125 ns, in this example the visually estimated delay is about 83 ns.
ETSI
119 ETSI TR 101 290 V1.2.1 (2001-05)
Figure E-23: Overall signal delay using a reference transmitter
Figure E-24: Direct measurement of the lobe's width, 1,3 MHz
Reference Transmitter
Transmitter undet test
TS
Generator
SFN
Adaptor
Test Signal
+ Spectrum
analyser
GPS
Receiver
1 pps
10 MHz
SFN
Adaptor
Modulator
RF
Stage
SFN
Adaptor
Modulator
RF
Stage
A.D.
Adjustable delay for the reference transmitter (Digital buffer)
ETSI
120 ETSI TR 101 290 V1.2.1 (2001-05)
Figure E-24 shows a lobe width of about 1,3 MHz, in a total span of 20 MHz or 2 MHz/division (the dual marker
facility was not set to this measurement, so a graphical approach was made), then the difference in delay between the
two transmitters is: D = 1/1,3 = 770 ns.
Figure E-25: Lobe's width wider than 8 MHz, (about 12 MHz)
Figure E-25 shows a lobe width, which may well be as wide as 12 MHz (visual estimation), in a total span of 20 MHz
or 2 MHz/division (the dual marker facility was not set to this measurement, so a graphical estimation was made), then
the difference in delay between the two transmitters is: D = 1/12 = 83 ns.
ETSI
121 ETSI TR 101 290 V1.2.1 (2001-05)
Annex F (informative):
Specification of test signals of DVB-T modulator
F.1 Introduction
In order to compare simulated data within a DVB-T modem it is necessary to specify test points, signal formats and a
subset of modes. The present document contains the specifications of how to do this. This specification should be
accurate enough to enable comparison of simulated data at different points within the modulator.
F.2 Input signal
Figure F-1: Input test sequence generator for DVB-T modulator
The number of bits in a super-frame is depending on the actual DVB-T mode. The maximum number of Reed-
Solomon/MPEG-2 packets in a super-frame is 5 292. This corresponds to 7 959 168 input bits that is shorter than a
maximum length sequence of length 223−1 = 8 388 607. The input test sequence to the modulator can therefore be
generated by a shift register of length 23 with suitable feedback. The generator polynomial should be 1 + x18 + x23. The
PRBS data on every 188 byte is replaced by the sync byte content, 47 HEX. This means that during the sync bytes the
PRBS generator should continue, but the source for the output is the sync byte generator instead of the PRBS generator.
The input test sequence starts with a sync byte as the first eight bits, and the initialization word in the PRBS generator is
"all ones". The PRBS generator is reset at the beginning of each super-frame. The test sequence at the beginning of each
super-frame starts with:
0100 0111 0000 0000 0011 1110 0000 0000 0000 1111 1111 1100 (first byte is sync byte 47 HEX).
The corresponding HEX numbers are: 47 00 3E 00 0F FC.
There are up to eight possible phases of the energy dispersal with respect to the start of the super-frame. The first sync
byte in the sequence, i.e. the first 8 bits should be inverted by the energy dispersal block. The length of the input signal
can in principle be arbitrary. However, it is not meaningful to have a sequence shorter than one OFDM symbol. The
maximum length will in practice be limited by the amount of data. Very large data files may be difficult to handle and
interchange. One super-frame is therefore regarded as the longest sequence of interest. The outer interleaver will spread
data across the super-frame boundaries. The ambiguity in the output sequence caused by this is circumvented by
using the second super-frame in the simulated sequence as the output signal. This means that the simulator should
produce one super-frame before useful data starts to appear at the output.
The file format for storing data allows for variable lengths of simulated data since the length indicator is contained in
the header of the file. Simulations with different lengths can therefore be compared over the length of the shortest
sequence.
ETSI
122 ETSI TR 101 290 V1.2.1 (2001-05)
F.3 Test modes
The file header in the test file contains information about the specific DVB-T mode used for the simulation. By reading
this information a complete description of the set-up is obtained. In order to ease comparison of data and to reduce the
amount of simulations necessary a set of "preferred modes" are defined. The preferred test mode for non-hierarchical
transmission is:
Inner code rate: 2/3;
Modulation method: 64 QAM;
FFT size: 8 k;
Guard interval: 1/32.
For hierarchical transmission the preferred mode is:
Inner code rate HP: 2/3;
Inner code rate LP: 3/4;
Modulation method: QPSK in 64 QAM, α = 2;
FFT size: 8 k;
Guard interval: 1/32.
F.4 Test points
The simulated data can be probed at different points within the modulator. Eight test points are defined, which are
related to the interfaces described in figure 9-1:
1) at input (A);
2) after mux adaptation, energy dispersal (B);
3) after outer encoder (C);
4) after outer interleaver (D);
5) after inner encoder (E);
6) after inner interleaver (F);
7) after frame adaptation (H);
8) after guard interval insertion (J).
F.5 File format for interchange of simulated data
The file header as well as simulated data from the modem are stored as ASCII characters on files with carriage return
and line feed at the end of each line. In order to interchange data it is important that the same file format be used by
everyone. A file containing such data should have a header which has the following information:
- text string with a maximum of 80 characters (affiliation, time, place etc.);
- "printf" string used to store the data in the data section of the file;
- test point description;
- lengthofdatabuffer;
ETSI
123 ETSI TR 101 290 V1.2.1 (2001-05)
- constellation;
- hierarchy;
- code rate (code rate for HP);
- code rate LP (Don't care for non-hierarchical modes);
- guard interval;
- transmission mode;
- simulated data (HEX or floating point).
The specification for each of these entries are given in tables F.1 to F.8.
F.5.1 Test point number
Table F.1: Test point number
Test point Interface Text contained in file header
1 A at input
2 B after MUX adaptation and energy dispersal
3 C after outer coder
4 D after outer interleaver
5 E after inner coder
6 F after inner interleaver
7 H after frame adaptation
8 J after guard interval insertion
F.5.2 Length of data buffer
The length indicator specifies the number of lines contained in the data section of the file which has two floating points
or one two digit HEX on each line.
F.5.3 Bit ordering after inner interleaver
The signal at test point 4 after inner interleaver should contain data from one carrier on each line. The bit ordering
should be according to table F.2.
Table F.2: Bit ordering in the signal representation at test point 4, after the inner interleaver
Modulation method Bit ordering Representation
QPSK y0q y1q 2-digit HEX (00 to 03)
16 QAM y0q y1q y2q y3q 2-digit HEX (00 to 0F)
64 QAM y0q y1q y2q y3q y4q y5q 2-digit HEX (00 to 3F)
F.5.4 Carrier allocation
The signal contains 1 705 or 6 817 active carriers for the 2 k and 8 k modes respectively. In order to ease comparison of
different data sets the allocation of these into the FFT bins should be specified. The signal is arranged such that it is
centred around half the sampling frequency.
ETSI
124 ETSI TR 101 290 V1.2.1 (2001-05)
Table F.3: Carrier allocation
FFT bins
containing zeros
FFT bins
containing active
FFT bins
containing zeros
2 k mode 0 to 171 172 (Kmin) to 1 876 (Kmax) 1 877 to 2 047
8 k mode 0 to 687 688 (Kmin) to 7 504 (Kmax) 7 505 to 8 191
F.5.5 Scaling
At test point 7 (after frame adaptation) the data should be scaled such that: "Vector length of a boosted pilot" is equal to
unity.
The gain factor through the IFFT should be equal to unity. This gain factor is defined as:
( )
( )
*
*


=
N
N
x x
z z
η
where x are the complex numbers representing one complete OFDM symbol at the input of the IFFT including data
carriers, pilots and null-carriers. And z is the complex signal for the corresponding OFDM symbol at the IFFT output
before guard interval insertion. The number N is equal to the IFFT size (2 k or 8 k). The asterisk denotes complex
conjugate. This ensures correct scaling of data at test point 8 (after guard interval insertion).
F.5.6 Constellation
The possible constellations are listed in table F.4. The file header should contain one of them.
Table F.4: Constellations
QPSK
16-QAM
64-QAM
F.5.7 Hierarchy
The hierarchical identifier specifies if hierarchical mode is on or off and also the alpha value in case hierarchical mode
is on. For non-hierarchical transmission alpha is set to one. Table F.5 contains the possible choices and the file header
should contain one of them.
Table F.5 Hierarchical identifier
Non-hierarchical, alpha = 1
Hierarchical, alpha = 1
Hierarchical, alpha = 2
Hierarchical, alpha = 4
F.5.8 Code rate LP and HP
The code rate identifiers specifies the code rate for the LP and HP streams. Table F.6 contains the possible choices and
the file header should contain one of them.
ETSI
125 ETSI TR 101 290 V1.2.1 (2001-05)
Table F.6: Code rate identifier
Code rate identifier
½
2/3
3/4
5/6
7/8
F.5.9 Guard interval
Table F.7 contains the possible choices for the guard interval and the file header should contain one of them.
Table F.7: Guard interval identifier
Guard interval identifier
1/32
1/16
1/8
1/4
F.5.10 Transmission mode
The transmission mode can be either 2 k or 8 k. Table F.8 contains the possible choices and the file header should
contain one of them.
Table F.8: Transmission mode identifier
Transmission mode identifier
2 048
8 192
F.5.11 Data format
The data at test point 1 to 6 arewritten to file using 2-digit HEXnumbers with "printf" string%X\n.
At test point 7 and 8 each line in the file contains real and imaginary parts with at least 6 significant decimal digits each.
The real and imaginary parts and separated by at least 2 spaces. The data is written to file using "printf" with % e\n.
F.5.12 Example
This is an example of a print-out of a file containing the data sequence at the input for the preferred mode for nonhierarchical
transmission. The text in parenthesis is just for explanation and should not be contained in the file.
Stockholm, May 22, 1996, example of input data. Preferred non-hierarchical mode:
%X\n (Data stored in HEX format);
at input (Data at test point 1 at modulator input);
758 016 (One super-frame of data);
64-QAM (Constellation 64 QAM);
non-hierarchical, alpha = 1 (Non hierarchical transmission);
2/3(2/3 inner code rate);
ETSI
126 ETSI TR 101 290 V1.2.1 (2001-05)
0 (Don't care. Code rate LP);
1/32 (Guard interval = 1/32);
8 192 (8 k IFFT size);
47 (First data byte is sync byte 47 HEX);
00 (Rest of data).
ETSI
127 ETSI TR 101 290 V1.2.1 (2001-05)
Annex G (informative):
Theoretical background information on measurement
techniques
This informative annex presents a review of the theoretical background to the measurement techniques recommended in
the present document. It is an attempt to gather the most relevant background information into one location, particularly
for the benefit of engineers and technicians who are new to digital modulation techniques. It is hoped that it will provide
a working knowledge of the theoretical and practical issues, particularly the potential sources of ambiguity and error, to
help users of the present document make valid, accurate and repeatable measurements.
G.1 Overview
The basic purpose of a digital transmission system is to transfer data from A to B with as few errors as possible. It
follows that the fundamental measure of system quality is the transmission error rate.
The transmission error rate is usually measured as the Bit Error Rate (BER), however it can also be informative to
consider the error rate of other transmission elements such as bytes, MPEG packets, or m-bit modulation symbols. In
practice, although a certain guaranteed minimum BER performance may be a system implementation goal, the system
BER alone is not a particularly informative measurement.
The most important figure of merit for any digital transmission system is the BER expressed as a function of the ratio of
wanted information power to unwanted interference power (C/N). This is underlined by the fact that most of the
measurements in the present document are built around this central theme of BER vs. C/N (or, equivalently, BER vs.
Eb/N0).
There are measurements of the individual elements (power and BER measurements). There are measurements of the
difference between theoretical and ideal performance (margin and degradation measurements). There are measurements
intended to help identify the sources of transmission errors (interference, spectrum, jitter and I/Q measurements). There
are measurements for monitoring the consequences of transmission errors at the system level (availability, error event
logging).
G.2 RF/IF power ("carrier")
When describing the Quadrature Amplitude Modulated (QAM) signals employed by DVB-C or the Quadrature Phase
Shift Keying (QPSK) signals employed by DVB-S, it is common to refer to the modulated RF/IF signal as "carrier" (C),
mainly to distinguish it from "signal" (S) which is generally used to refer to the baseband demodulated signal.
Strictly, it is incorrect to describe this signal as "carrier" because QAM and QPSK (which is equivalent to 4-state QAM)
are suppressed carrier modulation schemes. For OFDM, with thousands of suppressed carriers and assorted pilot tones,
the label "carrier" is even more inappropriate. This is why deliberately the expression "wanted information power" is
used in the paragraph above, and why the parameter is referred to as "RF/IF power" in the present document.
However, it is clear that engineers will continue to use "carrier" as a convenient shorthand for this parameter,
particularly when talking about the "carrier"-to-noise ratio. It seems futile to attempt to change this, so instead it is
clearly defined what is meant by "carrier" in this context. Carrier, more accurately called RF/IF power, is the total
power of the modulated RF/IF signal as would be measured by a thermal power sensor in the absence of any other
signals (including noise).
For DVB compliant systems the QAM/QPSK passband spectrum is shaped by root raised cosine filtering with a roll-off
factor alpha (α) of 0,15 for DVB-C systems, or 0,35 for DVB-S systems. For an ideal QAM/QPSK system this means
that all the RF/IF power will lie in the frequency band:
( )
2
( ) 1
S
OCC QAM C
f
BW = f ± +α × (G.1)
ETSI
128 ETSI TR 101 290 V1.2.1 (2001-05)
Equation G.1 defines the occupied bandwidth of the signal, where fC is the carrier frequency, fS is the symbol rate of
the modulation, and α is the filter roll-off factor. RF/IF power (or "carrier") is the total power in this "rectangular"
bandwidth, that is, with no further filtering applied.
For OFDM systems the definition of occupied bandwidth is expressed differently because of the radically different
modulation technique, however the principle is very similar. The OFDM "shoulders" are not considered to be wanted
information power, and are not included in the RF/IF power calculation, even though the power does actually come out
of the transmitter:
BWOCC(OFDM ) = n× fSPACING (G.2)
where n = 6 817 (8 k mode) or 1 705 (2 k mode) and fSPACING = 1 116 Hz (8 k mode) or 4 464Hz (2 k mode).
In a real multi-signal system (e.g. a live CATV network) measurement of the RF/IF power in a single channel requires a
frequency selective technique. This could employ a thermal power metre preceded by a suitably calibrated channel
filter, a spectrum analyser with band power measurement capability, or a measuring receiver. Depending on the
measurement technique a filter may be required to exclude the "shoulders" of a single OFDM signal.
G.3 Noise level
The noise level is the unwanted interference power present in the system when the wanted information power is
removed. This is a less bounded quantity than the RF/IF power because there is no definitively "correct" bandwidth
over which to measure the noise. The choice is to some extent arbitrary, but the "top three" choices are probably:
1) Channel bandwidth: In a channel based system such as a CATV network you could choose the channel
bandwidth, for example 8 MHz, as the system noise bandwidth. This is considered by the DVB-MG to be
inappropriate for C/N measurements in digital TV systems. It will result in misleadingly poor C/N ratios when
the modulation symbol rate is low relative to the available channel bandwidth. It unnecessarily complicates
conversion between C/N measurements made "in the channel " and "in the receiver " by introducing symbol rate
dependent correction factors.
2) Symbol rate: For digital modulation employing Nyquist filtering split equally between the transmitter and
receiver, the noise bandwidth of the receiver equals the symbol rate. This is considered by the DVB-MG to be
appropriate for "in the receiver " C/N measurements of digital TV systems since this reflects the amount of noise
entering the receiver independent of symbol rate.
3) The occupied bandwidth: For digital modulation employing Nyquist filtering the occupied bandwidth of the
modulated signal is (1 + α) × fS. This is considered by the DVB-MG to be appropriate for "in the channel " C/N
measurements of digital TV systems since it exactly covers the transmitted spectrum, independent of symbol
rate.
The DVB-MG have chosen occupied bandwidth, as defined by equation G.1, as the standard definition of noise
bandwidth in DVB-C and DVB-S systems. This is primarily because "in the channel " C/N is considered to be the
fundamental measurement, but also because a simple correction factor can be applied to determine the equivalent "in
the receiver " C/N value.
The other possibility that should be mentioned is to assume that the noise power is evenly distributed across the
frequency spectrum of interest and so can be described by a single noise power density value (N0) which is the noise
power present in a 1 Hz bandwidth. From this, the noise power present in any given system noise power bandwidth
(BWSYS) can be obtained by simple multiplication:
N = N0 × BWSYS (G.3)
By talking in terms of N0 we are freed from the need to define a noise bandwidth, but we are making an assumption that
the noise power spectrum is flat across the bandwidth of interest.
ETSI
129 ETSI TR 101 290 V1.2.1 (2001-05)
G.4 Energy-per-bit (Eb)
Trying to commission a DVB system against tight deadlines, Energy-per-bit (Eb) seems to be a rather academic
concept, particularly since the directly measurable quantity is RF power.
However, it is useful to understand Eb, even if only to avoid confusion when it appears in technical specifications or
discussions. Historically, use of Eb arises from information theory and as part of an academic desire to normalize the
performance of different modulation formats and coding schemes for comparative purposes.
The Energy-per-bit is the energy expended in transmitting each single bit of information. Eb is of little practical use on
its own, it is most useful in the context of a graph of BER vs. the Eb/N0 ratio - the well known "waterfall curve" (see
figures G-1 and G-2).
By normalizing to an Eb/N0 ratio on the X axis, the relative performance of various complexities of digital modulation
and channel coding can be compared because the scaling effects of actual signal and noise powers, number of
bits-per-symbol and symbol rate are removed. It is then simply a case of comparing the bit error probability for a given
ratio.
Energy-per-bit can be easily translated to carrier power. Power is energy-per-second. Which can be expanded to
energy-per-bit, times bits-per-symbol, times symbols-per-second. Expressed algebraically we get:
( ) C = Eb ×log2 M × fS (G.4)
G.5 C/N ratio and Eb/No ratio
The parameters that can be directly measured are RF/IF or "carrier" power (C) and noise power in a certain bandwidth
(N). From these measurements we can immediately compute the C/N ratio.
With the equations above, knowledge of the other parameters (e.g. fS ) and a little algebra we can also arrive at an
equivalent Eb/N0 ratio.
G.6 Practical application of the measurements
At this point it seems that C/N(or Eb/N0) is defined, and indeed it is from an algebraic perspective.
However, there is scope for endless confusion in applying these simple formulae unless the user is very clear about
where the C/N or Eb/N0 ratio is being measured, and what values are being used for the subordinate parameters, most
particularly the system noise bandwidth.
C/N (or Eb/N0) can be measured "in the channel" or "in the receiver". The meaning of "in the channel" is fairly selfevident,
"in the receiver" may need further explanation.
There are typically three filtering processes present in a receiver. The first (which is optional) is a relatively broadband
tuneable pre-selection simply to reduce the power presented to the receiver RF front-end. The second, usually applied at
an IF, is a high-order bandpass channel selection filter to extract the desired signal with (ideally) no modification of the
signal spectrum. The third is the root-raised cosine Nyquist filtering, commonly implemented in the low pass filters
following the I/Q demodulation.
For theoretical simplicity we assume that the receiver's bandwidth and band shape are defined totally by the low-pass
root-raised cosine filters because the intended purpose of the other RF/IF filters is only signal pre-selection. So we can
model the receiver as a broadband receiver with a root-raised cosine passband filter followed by I/Q demodulation.
With this in mind, "in the receiver" can be seen to mean "after the bandwidth and band shape modifying effects of the
receiver Nyquist filters has been taken into account".
Whether artificially generating a specific C/N ratio or just measuring the existing C/N ratio it is important to understand
the difference between the "in the channel" and "in the receiver" nodes.
ETSI
130 ETSI TR 101 290 V1.2.1 (2001-05)
On a more practical note, graphing the BER performance of a receiver versus Eb/N0 removes the ambiguity introduced
by varying noise bandwidth. If we use the "in the channel" Eb value then we get a certain BER curve, ifwe use the
slightly lower "in the receiver" Eb value then the Eb/N0 ratio is slightly poorer for the same BER, the curve moves to
the left (closer to the theoretical curve) and the implementation loss decreases because the loss due to the receive filters
is not included. An example may help to explain this.
G.7 Example
Creation of a signal with a specific C/N ratio in order to test the performance of an Integrated Receiver Decoder (IRD),
or perhaps to degrade an incoming RF/IF signal to a specific C/N ratio in order to establish the noise margin.
To do this, add broadband white Gaussian noise "in the channel " to the relatively noise free RF/IF signal. Measure (or
compute) the carrier power and then adjust the noise power density to give the required noise power in the selected
noise power bandwidth.
Taking the following QAM system parameters as an example:
Symbol rate: fS = 6,875MHz;
Filter roll-off: α = 0,15;
System noise bandwidth: BWNOISE = 8MHz;
Constellation size: M = 64;
Carrier power (in dB): C = -25 dBm.
then:
C = −25 dBm
Eb = C −10×log10 (log2 (M)× f S ) = −101,15 dBm
If a C/N ratio of 23 dB is wanted, then:
00 , 48 − =




= −
N dB
C
N C dBm
N0 = N −10× log10 (BWNOISE ) = −118,03 dBm
So the ratio of Carrier-to-Noise applied in an 8 MHz system bandwidth at RF/IF can be described as:
= 23,00
N
C
dB
16,88
0
=
N
Eb dB
This signal is then passed through the receiver root-raised cosine filters. The equivalent noise bandwidth of a bandpass
root-raised cosine filter is equal to the symbol rate fS. The noise power originally defined in an 8 MHz system
bandwidth is reduced accordingly:
66 , 48 log 10 10 − =






= + ×
NOISE
S
REC BW
f
N N dB
The noise power density N0 is unchanged by the receive filter:
N0(REC) = N0 = -118,03 dBm.
ETSI
131 ETSI TR 101 290 V1.2.1 (2001-05)
The signal power is already root-raised cosine shaped by the transmitter and so its power is only modified by the factor
(1-α/4):
25,17
4
1 log 10 10 − =




= + × −α
CREC C dB (G.6)
The Energy-per-bit Eb is subject to this same reduction factor: Eb(REC) = -101,32 dBm.
So the ratio of Carrier-to-Noise inside the receiver can be described as:
= 23,49
REC
REC
N
C
dB
16,71
0( )
( ) =
REC
b REC
N
E
dB
It is this received C/N (or Eb/N0) ratio that, when demodulated translates directly to a Signal-to-Noise Ratio (SNR) in
the I/Q domain. In the idealized case that white Gaussian noise is the only impairment present then this also determines
the Modulation Error Ratio (MER).
We can easily derive a general formula for the C/N modification due to the receive filters;


















−
= + ×
NOISE
REC S
REC
BW
N f
C
N
C 4
1
10 log10
α
dB (G.7)
and another for Eb/N0;




= + × −
4
10 log10 1
0( ) 0
( ) α
N
E
N
E b
REC
b REC dB (G.8)
For the C/N case the correction factor is dependent on filter roll-off, symbol rate and the system noise bandwidth used
to define the noise power. However, if the occupied bandwidth is used as the system noise bandwidth, then equation
G.7 simplifies to;











+





−
= + ×
α
α
1
1
4
1
10 log10 N
C
N
C
REC
REC dB (G.9)
and the correction factor becomes a constant dependent on the filter α only.
For DVB-C with filter α = 0,15 = + 0,441
N
C
N
C
REC
REC dB;
For DVB-S with filter α = 0,35 = + 0,906
N
C
N
C
REC
REC dB.
ETSI
132 ETSI TR 101 290 V1.2.1 (2001-05)
For comparison, if one were to always use the channel bandwidth (e.g. 8 MHz) as the system noise bandwidth then one
should use equation G.7, the correction factor becomes symbol rate dependent, and ranges from +0,441 dB for a
theoretical maximum occupancy symbol rate of 6,957 MBaud, through +0,492 dB for the example symbol rate of
6,875 MBaud, to +1,285 dB for a typical lower rate of 5,728 MBaud.
For the Eb/N0 case the correction for the DVB-C standard filter roll-off of α = 0,15 the correction factor is -0,166 dB,
and for the DVB-S standard filter roll-off of α = 0,35 it is -0,398 dB.
It is perhaps worth mentioning that using the C/N correction formula (equation G.7) gives correction factors which
suggest that the C/N is actually improved by the receive filter, but this is only because the system noise bandwidth is
larger than the receiver noise bandwidth.
The Eb/N0 formula (equation G.8) more accurately reflects reality, the information-to-noise ratio is actually degraded
by a small amount by the receive filter, because for the filter to pass the RF signal spectrum properly at the band edges
it should also pass proportionately more noise power than signal power.
G.8 Signal-to-Noise Ratio (SNR) and Modulation Error
Ratio (MER)
When a randomly modulated QAM or QPSK carrier and the associated passband noise is demodulated, approximately
half the signal power and half the noise power will be delivered into each baseband component channel (I and Q). The
demodulation process will have a certain gain, but this gain factor will apply equally to the signal and to the noise so the
resulting SNR in each channel will be approximately the same as the CREC/NREC ratio computed above.
The vector sum of the mean I and Q signal powers ratioed to the vector sum of the mean I and Q noise powers will, at
least theoretically, be exactly the same as the CREC/NREC ratio computed above.
This ratio of I/Q signal power to I/Q noise power expressed in dB is the definition given in the present document for
both SNR and for MER. The difference between these two measurements lines in what perturbations of the received
signal are included in the computation.
When the only significant impairment is noise then SNR and MER are equivalent, and are numerically equal to
CREC/NREC. The relationship between CREC/NREC and C/N depends on the choice of system noise bandwidth. If the
symbol rate is chosen as the system noise bandwidth (as defined in the present document clause 6.7) then the
relationship is a fixed offset of a fraction of 1 dB as described above.
This would appear to suggest that C/N measured in the passband can be equated directly to SNR in baseband.
Unfortunately other factors should also be considered in a real system. The SNR of the source modulator, the signal
amplitude dependence of the noise floor of system components, and the fact that the receiver equalizer will have the
effect of translating some linear impairments into noise. The exact interrelation of these parameters is the subject of
further study.
G.9 BER vs. C/N
As was stated in the introduction, the Bit Error Rate (BER) as a function of Carrier-to-Noise ratio (C/N) is the most
important figure of merit for any digital transmission system.
To evaluate the performance of modulator and demodulator realizations, measured BER values are compared against
the theoretical limits of the Bit Error Probability (BEP) PB. Regarding DVB satellite and cable transmission schemes
the BEP is usually determined based on the following assumptions:
- the only noise present is additive white Gaussian noise;
- the channel itself does not introduce any linear or non-linear distortions;
- modulator and demodulator are perfect devices (no timing errors, ideal band-limiting filters).
Based on these assumptions it is possible to calculate fairly accurate upper limits for BEP vs. C/N.
ETSI
133 ETSI TR 101 290 V1.2.1 (2001-05)
Since C/N depends on noise bandwidth it is common practice to normalize C/N by using Eb/N0 instead, where Eb is the
Energy-per-bit and N0 is the noise density. The transition from one value to the other is given by:
f m
BW
N
C
N
E
S
b NOISE
×
= ×
0
(G.10)
where BWNOISE is the equivalent noise bandwidth, fS is the symbol rate, and m is the number of bits-per-symbol,
m = log2(M), where M is the number of constellation points. When applying this formula it is important to be consistent
in using either the "in the channel" C/N or the "in the receiver" C/N values.
If Forward Error Correction (FEC) is employed, the information rate RI is increased up to the transmission rate RT by
adding the FEC information. The relation:
T
I
C R
R
R = (G.11)
is called the FEC rate. The transmission rate of an FEC rate 1/2 system for example will be 2 times the information rate.
Therefore the "Transmission Rate" Eb/N0 will be 3 dB less than the "Information Rate" Eb/N0, provided C/N stays
constant. This results from the fact that half of the available signal power is spent on FEC information. To compensate
for this effect Eb/N0 should be increased by 3 dB in case of "Information Rate" BEP. In general, if the BEP should be
calculated based on the information rate, Eb/N0 should be increased by 10 × log10(1/RC) dB.
If the performance of different FEC schemes is to be compared for power limited channels like satellite transmission,
the information rate should be used because it explicitly takes into account the signal power which is used for
redundancy only, and which is therefore lost for the information itself. In case of bandwidth limited channels like cable
results based on the transmission rate may be more appropriate.
G.10 Error probability of Quadrature Amplitude Modulation
(QAM)
Each state in an M state QAM constellation represents a log2(M) = m bit symbol. For example, each state in a 64 QAM
constellation represents a 6-bit symbol.
When the received signal is perturbed by Additive White Gaussian Noise (AWGN) there is a probability that any
particular symbol will be wrongly decoded into one of the adjacent symbols. The Symbol Error Probability PS of QAM
with M constellation points, arranged in a rectangular set, for m even, is given by (see bibliography: Proakis, John G.:
"Digital Communication", McGraw Hill, 1989):
( )
( )
( )
( )















×
× −
×
×





× − × −








×
× −
×
×





− × =





0
2
0
2
0 2 1
3 log
erfc
1
1
2
1
1
2 1
3 log
erfc
1
2 1
N
E
M
M
N M
E
M
M
N M
E
P b b b
S (G.12)
where erfc(x) is the complimentary error function given by:
( )
∞
= −
x
x e t dt
2 2
erfc
π
For practical purposes equation G.12 can be simplified by omitting the, generally insignificant, joint probability term to
give the approximation;
( )
( )






×
× −
×
×




 

− × =








0
2
0 2 1
1 3 log
2 1
N
E
M
M
erfc
N M
E
P b b
S (G.13)
This approximation introduces an error which increases with degrading Eb/N0, but is still less than 0,1 dB for 64 QAM
at Eb/N0 = 10 dB.
ETSI
134 ETSI TR 101 290 V1.2.1 (2001-05)
When M is not an even number (for example M = 5 (32 QAM) or M = 7 (128 QAM), then equation G.14 provides a
good approximation to the upper bound on PS (see bibliography: Proakis, John G.: "Digital Communication", McGraw
Hill, 1989):
( )
( )
2
0
2
0 2 1
3 log
1 1 erfc














×
× −
≤ − − ×







N
E
M
M
N
E
P b b
S (G.14)
As already stated, the above equations for Symbol Error Probability are based certain simplifying assumptions which
can be summarized as "the system is perfect except for the presence of additive white Gaussian noise", but within this
rather generous constraint the equations for PS are exact.
The corresponding Bit Error Probability (BEP) is less easily determined. It is directly related to the Symbol Error
Probability (SEP) but the exact relationship depends on how many bit errors are caused by each symbol error, and that
in turn depends on the constellation mapping and the use of differential encoding.
Two different approaches can be found in the literature. The first one makes no assumption about the constellation
mapping and is based on the probability that any particular bit in a symbol of p bits is in error, given that the symbol
itself is in error (see bibliography: Proakis, John G.: "Digital Communication", McGraw Hill, 1989 and see also Pratt,
Timothy and Bostian, Charles W.: "Satellite Communications", John Wiley & Sons, 1986). This approach leads to:
( )
p S
p
PB × P
−
=
−
2 1
2 1
(G.15)
The other approach assumes that an erroneous symbol contains just one bit in error. This assumption is valid as long as
a Gray coded mapping is used and the BER is not too high. Under these assumptions:
B PS
p
P = × 1
(G.16)
These approaches give different results for symbols of two or more bits. The second approach is generally adopted
because DVB systems employ Gray code mapping. The results tabulated in annex D are based on equations G.12 and
G.16.
It should be mentioned that for QAM systems DVB only employs Gray coding within each quadrant, the quadrant
boundaries are not Gray coded, and the mapping is partially differentially coded. Further work is required to establish
the exact PB to PS relationship for this combination of mapping and coding.
G.11 Error probability of QPSK
QPSK can be analysed as 4 QAM. Evaluation of the general QAM equation (G.12) for M = 4 gives:













× − ×








=







0 0 0
erfc
4
1
erfc 1
N
E
N
E
N
E
P b b b
S (G.17)
Again this can be simplified by dropping the joint probability term to give:






=





0 0
erfc
N
E
N
E
P b b
S
ETSI
135 ETSI TR 101 290 V1.2.1 (2001-05)
Using the PS to PB relationship defined in equation G.16, the expression for PB for QPSK modulation becomes:






× =





0 0
erfc
2
1
N
E
N
E
P b b
B (G.18)
G.12 Error probability after Viterbi decoding
Since it is not possible to derive exact theoretical expressions for the performance of convolutional codes, only upper
bounds can be presented in this annex. The upper bound:
( )






× × × × × ≤






∞
0 = 0
erfc
2
1 1
N
E
w d R d
N k
E
P b
c
d d
b
B
f
(G.19)
provides a good approximation for infinite precision, soft decision Viterbi decoding and infinite path history, as long as
Eb/N0 is not too low (see bibliography: Begin G., Haccoun D. and Chantal P.:"High-Rate Punctured Convolutional
Codes for Viterbi and Sequential Decoding", IEEE Trans. Commun., vol 37, pp 1113-1125, Nov. 1989 and also see
Begin G., Haccoun D. and Chantal P.: "Further Results on High-Rate Punctured Convolutional Codes for Viterbi and
Sequential Decoding", IEEE Trans. Commun., vol 38, pp1922-1928, Nov. 1990).
In equation G.19, df specifies the free distance of the used code, w(d) can be derived from the transfer function of the
convolutional code or determined directly by exhaustive search in the trellis diagram of the code, Rc= k/n is the rate of
the convolutional code, and Eb/N0 is given for the transmission rate. Since erfc(x) converges to zero quite quickly for
increasing x only very few terms of the sum should be taken into account. Values for df and w(d) can be found in
table G.1 regarding convolutional codes used in DVB satellite transmissions. The performance of convolutional codes
for low Eb/N0 values can only be evaluated by simulations.
Table G.1: Free distance and weights w(d) for DVB convolutional codes
Code Rate
Rc
1/2 2/3 3/4 5/6 7/8
free distance df 10 6 5 4 3
w(df) 36 3 42 92 9
w(df+1) 0 70 201 528 500
w(df+2) 211 285 1 492 8 694 7 437
w(df+3) 0 1 276 10 469 79 453 105 707
w(df+4) 1 404 6 160 62 935 791 795 1 402 089
w(df+5) 0 27 128 379 546 7 369 828 17 888 043
w(df+6) 11 633 117 019 2 252 394 67 809 347 221 889 258
w(df+7) 0 498 835 13 064 540 609 896 348 2 699 950 506
w(df+8) 2 103 480 75 080 308 5 416 272 113 32 328 278 848
w(df+9) 8 781 268 427 474 864 47 544 404 956 382 413 392 069
G.13 Error probability after RS decoding
A Reed-Solomon (RS) code is specified by the number of transmitted symbols (note) in a block N and the number of
information symbols K (see bibliography: Odenwalder J.P.: "Error Control Coding Handbook", Final report prepared
for United States Airforce under Contract No. F44620-76-C-0056, 1976).
Such a code will be able to correct up to t = (N-K)/2 symbol errors. As for DVB transmission N = 204 and K = 188 are
used. Therefore up to t = 8 erroneous symbols can be corrected.
ETSI
136 ETSI TR 101 290 V1.2.1 (2001-05)
NOTE: Whereas the symbols mentioned in context with QAM and QPSK are related to the modulation the
symbols mentioned here are just a group of bits.
The probability PBLOCK of an undetected error for a block of N symbols as a function of the error probability of the
incoming symbols PSIN is given by:
( )
+ =
× × − −





=
N
i t
N i
SIN
i
BLOCK PSIN P
i
N
P
1
1 (G.20)
From this expression the probability:
( )N i
SIN
i
SIN
N
i t
S i P P
i
N
N
P −
= +
− × ×





= × × 1
1
1
β (G.21)
of a symbol error can be derived, where βi is the average number of symbol errors remaining in the received block
given that the channel caused i symbol errors. Of course βi = 0 for i ≤ t. When i > t, βi can be bounded by considering
that if more than "t" errors occur, a decoder which can correct a maximum of "t" errors will at best correct "t" of the
errors and at worst add "t" errors. So:
i − t ≤ β i ≤ i + t (G.22)
is the possible range for β
i. A good approximation is β
i = i but also β
i = t + i is used, which can be regarded as an upper
limit. From G.21 the BEP can be calculated by using G.15 or G.16.
G.14 BEP vs. C/N for DVB cable transmission
For DVB transmission in cable networks, QAM-M systems with M = 16, 32 and 64 are specified. To evaluate the BEP
after RS decoding, the following steps should be done:
a) calculate the SEP after QAM demodulation by using (G.12) or (G.14);
b) transform the SEP into a BEP by applying (G.15) or (G.16) to the SEP with p = m;
c) transform the resulting BEP into a SEP with p = 8 by using (G.15) or (G.16);
d) use (G.21) to calculate the SEP PS after RS decoding;
e) apply (G.15) or (G.16) to PS with p = 8 to determine the final BEP;
f) if the BEP should be based on the information rate, shift the curve by:
- 10× log10(204/188) = 0,35 dB to the right.
If just the BEP before Reed-Solomon is needed, only the first two steps are necessary. In this case there is no difference
between information rate and transmission rate. All bits are regarded as information bits.
The limits before and after Reed-Solomon decoding for M = 64, β
i = i and Eb, based on the transmission rate, are
presented in figure G-1.
ETSI
137 ETSI TR 101 290 V1.2.1 (2001-05)
64-QAMDemodulation and Reed Solomon Decoding
1E-13
1E-11
1E-09
1E-07
1E-05
1E-03
13 15 17 19 21 23
Eb/N0 [dB]
Bit Error Rate (BER)
QAM Demodulation
Reed Solomon Decoding
Figure G-1: BER for QAM-64 DVB cable transmission
G.15 BER vs. C/N for DVB satellite transmission
For satellite transmission three different BEPs are possible:
- BEP after QPSK demodulation;
- BEP after Viterbi decoding;
- BEP after Reed-Solomon decoding.
The BEP after QPSK can be derived from (G.17). There is no difference to be made between information bit rate and
transmission bit rate.
The BEP after Viterbi decoding is expressed by (G.18). The result is based on the information rate, because RC is taken
explicitly into account in (G.18).
BEP after Reed-Solomon decoding can be derived from the above result by applying the following steps to the outcome
of (G.18):
a) transform the BEP after Viterbi decoding into a SEP by using (G.15) or (G.16) with p = 8;
b) use (G.17) to determine the SEP after Reed-Solomon decoding;
c) apply (G.15) or (G.16) to PS with p = 8 to determine the final BEP;
d) if the BEP should be based on the information rate, shift the curve by:
10 × log10(204/188) = 0,35 dB to the right.
The results for the three different BEPs and for all the different code rates Rc are presented in figure G-2.
ETSI
138 ETSI TR 101 290 V1.2.1 (2001-05)
QPSK Demodulation, Viterbi and Reed Solomon Decoding
1E-12
1E-10
1E-08
1E-06
1E-04
1E-02
0 2 4 6 8 10 12 14
Eb/N0 [dB]
Bit Error Rate (BER)
1/2
2/3
3/4
7/8
5/6
QPSK-Demodulation
Viterbi
Reed-
Solomon 1/2
2/3
3/4
5/6 7/8
Figure G-2: BER for DVB satellite transmission
Since it is common practice in satellite transmission to refer the results to the information rates the curves for BEP after
Reed-Solomon decoding have been shifted accordingly. The expression (G.19) is only valid for low error rates. Despite
the fact that for decreasing Eb/N0 the BER should converge to 1/2 the results according to (G.19) will posses a
singularity for Eb/N0 = 0. This behaviour is especially pronounced for Rc = 7/8, where the assumption of a low error
rate is not fulfilled above a BEP of 10-4.
G.16 Adding noise to a noisy signal
In a practical situation where we deliberately add noise to real signal in order to create a specific C/N ratio for
measurement purposes, it is important to realize that there are two fundamental assumptions implicit in this technique.
The first assumption is that the input signal has a high C/N ratio and can, for practical purposes, be regarded as carrier
only. The second assumption is that the input signal has a considerably better C/N ratio than the C/N ratio we wish to
generate. In practice we may be adding noise to an already noisy signal, and in this case there are accuracy issues
related to the above assumptions that should be considered.
First consider how noise is typically added to a signal. Figure G-3 gives a simplified block diagram.
ETSI
139 ETSI TR 101 290 V1.2.1 (2001-05)
Figure G-3: Simplified block diagram of C/N test set
The input is the carrier signal to be impaired. The carrier power is measured using the power metre. A broadband
Gaussian noise source is then filtered and attenuated appropriately to deliver the required noise density (N0) across the
frequency band of interest. The same power metre is used to set the noise power which helps ensure good C/N0 ratio
accuracy, The generated noise is added to the input signal to achieve the required C/N0 ratio in the output signal.
Finally, the carrier power is monitored and the power of the noise source is adjusted accordingly to maintain the
required C/N0.
In automated versions of this process, the user simply selects the desired C/N0 ratio. This can be entered as C/N0, but it
is more typically entered as C/N which requires that the user also enters the receiver or system noise bandwidth, or it
can be input as Eb/N0 which requires that the user also enters the system bit rate.
From this description it is evident that it is assumed that all the measured input power is carrier and the noise power to
achieve the required C/N ratio is computed accordingly. If the input already contains some noise or other carriers then
this will:
a) appear at the output in addition to the generated noise;
b) cause the generated noise power to be too large because it is based on the C + N power at the input, not just the
C power. This error is exacerbated if the input is not band limited.
We can derive a formula for the actual output C/N ratio as a sum of the theoretical C/N ratio and an error term:
1442443 14444244443
errorterm
N N N
N
C Nratio
theoretical
N
C
CN
i c n
c
c
actual




+ +
× −




= × 10 10 log10
/
10 log dB (G.23)
Where Nc is the noise power added due to the carrier power, Ni is the noise power already present in the input, Nn is the
noise power added due to the input noise. If we perform further manipulation of the error term then we arrive at an
expression in terms of the fractional input and output C/N ratios.





+ +
= ×
1
1
1
10 log10
in
out
in
error
CN
CN
CN
CN dB (G.24)
The error becomes significant if either the 1/CNin or the CNout/CNin term in the denominator moves away from zero
which will happen if either the C/Nin ratio or the C/Nout to C/Nin margin is reduced.
ETSI
140 ETSI TR 101 290 V1.2.1 (2001-05)
The present document gives a minimum value of 15 dB for the C/Nin ratio and for the C/Nout to C/Nin margin as a
guideline figure. To meet this condition in satellite systems it is necessary to use a sufficiently large dish to get the
required C/N ratio. A received C/N ratio of 20 dB or more is desirable.
Alternatively, it is possible to work with higher noise signals if it is possible to measure the carrier and noise power
accurately, for example by measuring carrier plus noise then switching off the carrier and measuring noise only.
Equation G.23 can then be used to compensate for the errors due to the input noise.
ETSI
141 ETSI TR 101 290 V1.2.1 (2001-05)
Annex H: Void
ETSI
142 ETSI TR 101 290 V1.2.1 (2001-05)
Annex I (informative):
PCR related measurements
This annex provides background information on the concept of PCR related measurements and the reasoning behind the
definition of the parameters in clause 5.3.2.
The aim is to gather the information which enables different implementations of PCR related measurements to show
consistent and comparable results for the same Transport Stream.
I.1 Introduction
Recovering the 27 MHz clock at the decoder side of a digital TV transmission system is necessary to re-create the video
signal. To allow recovery of the clock, the PCR values are sent within the Transport Stream. It is required that the PCR
values are correct at the point of origin and not distorted in the transmission chain to the point of creating problems in
the process of decoding the compressed signals.
Measuring the interval between arrival of PCR values, the accuracy of the expected values and the jitter accumulated on
those PCR values transmitted in a Transport Stream is necessary to assure the confidence of decodability of such
stream.
As jitter and drift rate are important parameters for the overall process, a clear definition is needed for what is
understood as PCR jitter and a guidance to its measurement method.
I.2 Limits
From the specifications set in ISO/IEC 13818-1 [1] it is possible to define a limit mask for the frequency deviation from
the nominal 27 MHz.
Frequency offset: is the difference between the actual value and the nominal frequency of the clock (27 MHz) . The
limit is set to ±810 Hz. Converting this value into relative or normalized units results in 810/27 × 106 = 30 × 10-6. This
means that the frequency of the clock at any moment should be the nominal ±0,003 %, or the nominal ±30 ppm. Rating
the limit of the frequency offset as relative has the advantage of obtaining a limit valid for any value of frequency for a
reference clock used to synthesize the nominal clock of 27 MHz. For example, the frequency error in Hz of a 270 MHz
serial clock derived from the 27 MHz system clock can be divided, or normalised, by 270 MHz to determine if the
frequency offset is within 30 ppm.
Frequency rate of change, or frequency drift rate: is the "speed" at which the frequency of a clock varies with time.
In other words it is the first derivative of the frequency with respect to time or the second derivative of phase with
respect to time.
The limit is set to 75 milli-hertz per second for the 27 MHz clock. It can be converted into relative limit by dividing by
27 MHz which produces a result of 75 × 10-3/27 × 106 = 2,777... × 10-9/s.
It means that the maximum rate of change allowed for the clock frequency is ±0,000 000 277 7...%/s of the nominal
value, or ±0,002 77...ppm/s of the nominal, or ±2,77...ppb/s of the nominal value of the system clock frequency. (Note
that a billion is taken here as 109, in many countries a billion is represented as 1012).
This result can also be presented as 0,001 %/hour, or as being 10 ppm/hr.
27 000 000 - 810 ≤ system_clock_frequency ≤ 27 000 000 + 810 @ 27 MHz
Frequency tolerance = ±30 × 10-6 @ 1 Hz (I-1)
Rate of change of system_clock_frequency ≤ 75 × 10-3 Hz/s@ 27 MHz
Drift tolerance = ±2,7778 × 10-9 /s@ 1 Hz (I-2)
ETSI
143 ETSI TR 101 290 V1.2.1 (2001-05)
Phase tolerance = ±500 × 10-9 s (I-3)
This represents the maximum error of a PCR value with respect to its time position in the Transport Stream.
The maximum limit for the phase represented in a PCR value is ±500 ns, this value is an absolute limit at the generation
of PCRs and does not include network-induced jitter.
The document ISO/IEC 13818-9 [3] (Extension for real time interface for systems decoders) specifies in clause 2.5
(Real-Time Interface for Low Jitter Applications a limit for t-jitter equal to 50 μs.
Low jitter applications tolerance = 25 × 10-6 s (I-3b)
NOTE: The limits for frequency offset and drift rate are imposed for the system clock as it is represented by the
values of the corresponding PCR fields. They include the effects of the system clock and any possible
errors in the PCR calculation. The limit of 500 ns is not imposed to the system clock, but to the accuracy
representing the PCR values with respect to their position in the Transport Stream. However the PCR
errors are fully equivalent to phase and jitter errors when the PCRs are used at the decoding point to
reconstruct the system clock.
I.3 Equations
The waveform of the phase modulation may have any shape that can be analysed as a composition of sinusoidal
waveforms of various amplitudes and phases. Also the clock may be a pulsed signal. In this case the formulas below
apply to the fundamental component of such periodic signal.
For example, the equation for a sinusoidal clock with sinusoidal phase modulation can be written as:
Fclk (t) = A × sin [ωc × t + Φ(t)] = A × sin [ωc × t + Φp × sin (ωm × t)]
where:
ωc nominal angular frequency of the program clock, (ωc = 2π × 27 MHz);
Φ(t) phase modulation function;
Φp peak phase deviation in radians;
ωm phase modulating angular frequency in units of radians/s.
The instantaneous phase of the clock has two terms as:
Φi(t) = ωc × t + Φ (t) = ωc × t + Φp × sin (ωm × t) (I-4)
The instantaneous angular frequency of the clock is found as the first derivative of the instantaneous phase as:
ωi (t) = d Φi (t)/d t = ωc + Φp × ωm × cos (ωm × t) (I-5)
where:
ωi instantaneous angular frequency of the clock, ωi = Φi
', in units of radians/s.
The frequency rate of change, or drift rate, is given by the first derivative of the angular frequency, or the second
derivative of the phase as:
ri (t) = d ωi (t)/d t = −Φp × ωm
2 × sin (ωm × t) (I-6)
where:
ri instantaneous rate of change of the clock, ri = Φi
'', in units of radians/s2.
ETSI
144 ETSI TR 101 290 V1.2.1 (2001-05)
I.4 Mask
A limit mask can be derived as a group of functions representing the limit specifications.
From the instantaneous phase equation (I-4) it can be seen that the maximum peak value of phase modulation is Φp
which can be compared to the limit set by ISO/IEC 13818-1 [1].
The phase equation may be found as:
Φp = ωc × Tmax = 2π × 27 MHz × 500 × 10-9 s = 84,823 radians (I-7)
where:
Tmax maximum time error of clock edge = 500 × 10-9 s
From the instantaneous angular frequency equation (I-5) it can be seen that the maximum peak value of angular
frequency offset is given by Φp× ωm which can be compared to the limit set by ISO/IEC 13818-1 [1] of 810 Hz.
The maximum angular frequency deviation from the nominal is:
Φp ×ωm = 2π × 810 radians/s
By dividing by ωm, the frequency equation for peak phase error as a function of modulation frequency may be found
as:
Φp = 2π × 810/ωm (I-8)
From the instantaneous drift rate equation (I-6) it can be seen that the maximum peak value of angular frequency
drift-rate is Φp · ωm
2
which can be compared to the limit set by ISO/IEC 13818-1 [1] of 75 mHz/s.
Φp × ωm
2 = 2π × 0,075 radians/s2
By dividing by ωm
2, the drift rate equation for peak phase error as a function of modulation frequency may be found as:
Φp = 2π × 0,075/ωm
2 (I-9)
All three equations may be normalized by dividing by 2π × 27 MHz.
The phase equation becomes:
Tmax = Φp/2π × 27 × 106 = 8,823/2π × 27 × 106= 500 × 10-9 (seconds) (I-7a)
The frequency equation becomes:
Tf(ωm) = Φp/2π × 27 × 106 = 2π × 810/(2π × 27 × 106 × ωm) = (30 × 10-6/ωm) s (I-8a)
The drift rate equation becomes:
Tr(ωm) = Φp/2π × 27 ×106 = 2π × 0,075/(2π × 27 × 106 × ωm
2) = (2,7778 × 10-9/ωm
2) s (I-9a)
The three equations (I-7a, I-8a and I-9a) can be seen in the graph of figure I-1.
ETSI
145 ETSI TR 101 290 V1.2.1 (2001-05)
Figure I-1: PCR jitter components
I.5 Break frequencies
Values for two break frequencies of figure I-1.
F1 can be found by re-arranging the equations for frequency and drift rate (I-8 and I-9 respectively) and solving for the
value of ωm that provides the same peak phase error:
Φp = 2π × 810/ωm and Φp = 2π × 0,075/ωm
2 radians
ωm = 2π × 0,075/2π × 810 = 9,2592 × 10-5 radians/s
F1 = ωm/2π = 14,736 × 10-6 Hz
The break frequency F1 is extremely low to have any practical use. When the frequency offset is to be measured there is
no need to wait about 5 days to have an averaged result appropriated to the period of such a signal. It is not considered
here due to its very long-term significance. It can be seen that the drift limit is enough for practical purposes of jitter
analysis.
F2 can be found by re-arranging and solving the equations of phase and drift rate (I-7 and I-9 respectively) for the value
of ωm that has the same peak phase error:
Φp = 84,823 radians and Φp = 2π × 0,075/ωm
2 radians
ωm = 0,4712 84,823 = 0,074535 radians/ s
F2 = 0,074535/2 = 0,01186 Hz
1 ,E-08
1 ,E-07
1 ,E-06
1 ,E-05
1 ,E-04
1 ,E-03
1 ,E-02
1 ,E-01
1 ,E+00
1 ,E+01
Frequency
of phase
variation,
in Hz
T in seconds
10-7 10-6 10-5 10-4 10-3 0,01 0,1 1 10 100 1000 F1 F2
Phase error Limit = 500 ns
Tmax = 500 · 10-9 (seconds)
Frequency Offset Limit = 810 Hz
Tf(ωm) = 30 · 10-6 / ωm (seconds)
Frequency Drift Limit = 0.075 Hz/s
Tr(ωm) = 2.7778 · 10-9 / ωm
2 (seconds)
Fig. I-1
ETSI
146 ETSI TR 101 290 V1.2.1 (2001-05)
NOTE: The same values may be obtained by using the normalized equations I-7a, I-8a and I-9a.
This break frequency (F2~10 mHz) is the recommended value by DVB-MG as the demarcation frequency for separating
the measurements of jitter and drift. It has been defined as filter MGF1 in the table 5.1.
This value defines the corner frequency to be used in the filters for processing the PCR data. A mask can be drawn from
the two equations used to obtain this value (phase equation I-7a and drift equation I-9a).
The mask so defined is represented in figure I-2.
Figure I-2: Mask for PCR jitter components
It can be seen that the maximum drift of 75 mHz/s may only be reasonably applied to jitter frequencies lower than the
demarcation frequency. Above such frequency it is possible in practice to find drifts much faster than the limit, when
real PCR errors are considered.
Above the demarcation frequency, the limit that applies is the absolute 500 ns for any PCR value.
NOTE: For the Low Jitter Applications (ISO/IEC 13818-9 [3]) the ±25 μs limit yields a demarcation frequency of
1,67 mHz, to be used in place of the 10 mHz. This suggests the use of a filter with about 2 mHz break
frequency when checking against this limit. This filter has been lumped under MGF4 due to the long time
constant involved, which makes it to provide a very slow response for a practical implementation.
I.6 Further implicit limitations
From figure I-2 it can be seen that a practical limit is also imposed to the ability to measure jitter frequencies above a
certain frequency.
For PCR values inserted at the minimum rate of 100 ms as per ISO/IEC 13818-1 [1] the samples arrive to the
measurement instrument at a 10 Hz rate. The Nyquist value (half the sampling rate) is equal to 5 Hz.
For PCR values inserted at the minimum rate of 40 ms as per TR 101 154 [4] the samples arrive to the measurement
instrument at a 25 Hz rate. The Nyquist value is equal to 12,5 Hz.
1 ,E-09
1 ,E-08
1 ,E-07
1 ,E-06
1 ,E-05
1 ,E-04
1 ,E-03
1 ,E-02
1 ,E-01
F2 = 0.01186 Hz
Demarcation
frequency
12.5 Hz
1/2 PCR repetition rate
(Minimum Nyquist
limit for DVB)
5 Hz
1/2 PCR repetition rate
(Minimum Nyquist
limit for MPEG-2)
Drift limiting region Jitter limiting region
T in seconds
10-5 10-4 10-3 0,01 0,1 1 10 100
Frequency
of phase
variation,
in Hz
Phase error Limit = 500 ns
Tmax = 500 · 10-9 (seconds)
Drift Limit = 0.075 Hz/s
Tr(ωm) = 2.7778 · 10-9 / ωm
2
Fig. I-2
Region where timing
errors exceed phaseerror
limit but do not
exceed drift-rate limit.
Region where timing
errors exceed drift-rate
limit but do not exceed
phase-error limit.
ETSI
147 ETSI TR 101 290 V1.2.1 (2001-05)
If higher PCR insertion rates are used in any of the above environments, the corresponding Nyquist frequency increases
proportionally. This implies that any statistics made by the measurement instrument based in jitter spectral analysis has
to measure the actual PCR rate.
Depending of the type of analysis, it is necessary to take in account that the PCR samples do not necessarily arrive at
regular intervals. For any practical implementation the designer may decide what is the preferred way for implementing
the filters: DSP techniques (IIR or FIR filters), with interpolation (linear, sinx/x, etc.) or without interpolation, analogue
circuitry or hybrid technology by mixing analogue and numerical analysis, etc.
It is interesting to note, however, that in most practical cases the rate of samples will occur at very high frequencies
(1000 times higher) compared to the frequency break points of the proposed filters (MGF1 at 10 mHz). The minimum
rate for PCRs is 10 Hz for general MPEG Transport Streams (25 Hz in DVB systems) and at this over-sampled PCR
values the transient response shape of filters with bandwidths near 10 mHz are not significantly affected by the
non-uniform rate.
I.7 Measurement procedures
It is possible to do jitter measurements fitting the data with a second-order curve (quadratic regression) limited by driftrate
specification. However, this is not necessary if one takes the view of creating separate measurements of jitter
and frequency-offset/drift-rate based on the more familiar method of sinusoidal spectral content of the timing
variations.
demarcation frequency
500ns
Drift-rate
Limit = 10ppm/Hr
jitter spec
2
1
1/2 sync-byte rate
Wander region Jitter region
Figure I-3: Total spectral mask of timing variations
For jitter spectral components below the demarcation frequency, the peak sinusoidal components of the PCR timingerror
can increase proportional to the square of the period of the spectral component without exceeding the drift-rate
limit of 10 ppm/Hr (also, equivalently, 2,8 ppb/s and 75 mHz/s @27 MHz). Since the decoder PLL and all subsequent
video timing equipment track this error, these components can far exceed the peak limit of 500 ns.
By inverting the specification mask, a spectrally weighted measurement or measurement filter becomes apparent as
follows in figure I.4.
ETSI
148 ETSI TR 101 290 V1.2.1 (2001-05)
demarcation frequency
0 dB
Total spectal weighting response combining
jitter and wander perturbations into one
measurement.
2
1
1/2 sync-byte rate
Wander region Jitter region
Figure I-4: Filter by inverting the spectral mask of timing variations
This can be decomposed into two separate measurements such that the sum of the Jitter and Drift-rate measured outputs
is essentially the same as the original.
demarcation frequency
0 dB
Jitter HPF
3
1
1/2 sync-byte rate
Wander region Jitter region
demarcation frequency
0 dB
Drift-rate measurement
response.
2
1
1/2 sync-byte rate
Wander region Jitter region
1
1
Frequency-offset
measurement response.
1
1
Figure I-5: 3rd order HPF for jitter and 1st order roll-off for drift measurements
Now jitter can be evaluated against given performance limits somewhat independently of the frequency drift-rate
performance limits. Note that in figure I-5 the Jitter HPF has a third-order response to reject the drift-rate components
from the measurement. Also in figure I-5 right, the Drift-rate measurement response has a first-order roll-off to reject
the jitter components from it's output. Also shown is the preferred Frequency-offset measurement response which, also
rejects jitter spectral components. Note (see figure I-5 right) that below the demarcation frequency, the
Frequency-offset is a first-derivative slope and the Drift-rate is a second-derivative slope.
The timing error need not be directly measured since it's time-derivative or frequency-offset contains all that is needed
to implement the measurement filters. This means that only two samples to compute the time-delta or
first-past-difference of the byte arrival time are needed. This is equivalent to measuring the instantaneous frequency
offset rather than the actual time-error of the transport stream and greatly simplifies the measurement with no loss in
information.
I.7.1 PCR_Accuracy (PCR_AC)
The result of PCR_AC is obtained at interface E of figure I-6.
The PCR_ACs that affect the PLL clock recovery for a specific program can be measured independently of arrival-time
by extracting the change in adjacent PCR values and the number of bytes between PCR's as follows:
K(i) = i' - i'', bytes, [PCR(i) - PCR(i-1)]/FNom - K(i)/TR = d(PCR_AC(i))/dt
TR = nominal Transport Stream rate, bytes/s, FNom = 27MHz
K(i) = number of bytes between current PCR(i) and previous PCR(i-1)
ETSI
149 ETSI TR 101 290 V1.2.1 (2001-05)
All high-pass and low-pass filter bandwidths as MGF1, MGF2, MGF3 and MGF4.
Figure I-6: PCR_Accuracy measurement
Note that this method measures PCR_AC independently of arrival-time. This can only be done for constant bitrate TS.
Drift-rate and frequency-offset are not measured. PCR interval errors are also not measured but can be determined
indirectly from K(i) /TR. Also note that PCR_AC is measured above the demarcation frequency to be consistent with
those spectral components that contribute to PLL jitter. The drift components of PCR_AC are likely negligible
compared to clock drift.
The second-order high-pass filter represents a second-order HPF response to the PCR accuracy due to the
first-derivative effect of the first-past-difference calculation of the PCR's shown in the diagram. This is best illustrated
as a discrete-time system operating at the average PCR rate in figure I.7.
(2)
HPF
2nd-order
2nd-order
cascade filter
1st-past
difference
Figure I-7: Second order HPF
In terms of the reference model presented in clause 5.3.2.1, diagram I-6 measures the difference in two PCR
inaccuracies Mp,i' – Mp,I''. A series of these measurements can be processed further to derive the individual PCR
inaccuracies Mp,I by assuming that average inaccuracy is zero.
I.7.2 PCR_drift_rate (PCR_DR)
The result of PCR_DR is obtained at interface H of figure I-8.
ETSI
150 ETSI TR 101 290 V1.2.1 (2001-05)
This measurement result is obtained after the combined action of the second order HPF represented by the loop (before
the integrator represented by the adder and latch), followed by the first order LPF. This combined action provides the
response indicated in figure I-5 for drift rate.
I.7.3 PCR_frequency_offset (PCR_FO)
The result of PCR_FO is obtained at interface G of figure I-8.
This measurement is obtained after the combined action of the first order HPF represented by the loop and the integrator
(represented by the adder and latch) followed by the first order LPF. This combined action provides the response
indicated in figure I-5 for frequency offset.
I.7.4 PCR_overall_jitter Measurement
The result of PCR_OJ is obtained at interface J of figure I-8.
This measurement result is obtained after the combined action of the second order HPF represented by the loop (before
the integrator represented by the adder and latch), followed by the first order HPF. This combined action provides the
response indicated in figure I-5 for jitter (left drawing).
Overall jitter includes the composite effect of PCR accuracy errors and PCR arrival-time jitter. It is important since this
relates directly to the effect on the program recovered clock jitter and drift. This method should also include a
measurement of clock drift-rate and frequency-offset. Therefore, the most practical method is to implement a SCF
recovery PLL like the one in the program decoder. By carefully controlling the bandwidth and calibrating the VCXO, it
is possible to measure, simultaneously, PCR overall jitter, SCF frequency-offset, and SCF drift-rate with the frequency
responses described before.
Figure I-8: Overall PCR jitter measurement combining the effects of
PCR_AC and PCR_arrival-time_jitter
ETSI
151 ETSI TR 101 290 V1.2.1 (2001-05)
Explanation:
Note that the PLL is a Type II control system with two ideal integrators (digital accumulator shown and VCXO). This
creates a 2nd –order high-pass closed-loop response at the output of the phase subtraction. Therefore, below the loop
bandwidth, the response is proportional to drift-rate and proportional to jitter above the loop bandwidth. It is necessary
to add an additional 1st-order HPF to the jitter measurement to remove the effects of drift-rate. Conversely, it is
necessary to add a 1st-order LPF to the drift-rate output to remove the effects of jitter from that measurement.
NOTE 1: If the filters are implemented using DSP techniques on the raw data, and since the PCR_rate is the sample
rate, the average PCR_rate should be determined by measuring the PCR_interval and filtering the result
with a 10 mHz LPF or lower. The value of PCR_rate can be used for those values shown in the figure to
effect the selected measurement bandwidth, BW, such that it is independent of PCR_rate.
NOTE 2: The design shown is a digital/analogue hybrid with a DAC driving the analogue loop filter. For a 14-bit
DAC the SF would be 2-14. The VCXO with gain Kv can be constructed from a sub-system consisting of
an OCXO and a FLL locking a VCXO. This can be used to calibrate the Frequency-offset output to the
wanted accuracy if desired. Otherwise, the VCXO can be used alone and its frequency error or offset
verified by applying a known, accurate frequency, TS and subtracting the error from subsequent
measurements.
NOTE 3: Alternatively, a free-running OCXO can be used to determine the PCR_interval with know methods and a
numerical VCO can be constructed. With this method a completely digital or software only version can be
constructed using the measured PCR_interval and the PCR values. It can be shown that this method can
have a bandwidth that is essentially independent of average PCR_rate with the measured jitter values
relatively independent of variations in PCR_interval.
Although this method describes a PLL implementation as a hybrid of DSP and analogue signal processing, other
methods that yield the same filtered responses are possible.
I.8 Considerations on performing PCR measurements
The measurement and validation of contributions to jitter and drift rate of a program STC carried by it's discrete-time
samples via PCR values of each program in a TS requires certain mathematical analysis of such samples in order to
compute the performance limits for direct comparison to those fixed in the standards.
Typical sampled system analysis relies on a regular sampling rate of the data to be analysed. This is not generally the
case of the discrete-time samples carried by PCR values which, per their own nature, depend on criteria and priorities at
the multiplexing stage.
The ITU-T Recommendation H.222.0/ISO/IEC 13818-1 [1] establishes a maximum interval of 100 ms between
consecutive PCR values. The DVB recommends that all DVB compliant systems will transmit the PCR values with a
maximum interval of 40 ms, but all receivers should work properly with intervals as long as 100 ms.
None of the standards forced that the interval, whatever it is, should be constant. This is because in the multiplexing
process there is a need for an allowance as to the instant the packet containing the PCR field for a given program is to
be inserted into the TS. However the intention of the designers and operators of multiplexers is to provide such values at
the most regular rate as possible.
At the receiver the regeneration of the 27 MHz of system clock for the program under the decoding process is
controlled by a signal that makes use of each of the PCR values corresponding to such program at the time of arrival to
introduce corrections when needed. It is assumed that the stability of the clock regenerator is such that the phase does
not unduly drift from one PCR value to the next over intervals as long as 100 ms.
However, it is the responsibility of the TS to provide the values of PCR correctly with an error no greater than 500 ns
from the instantaneous phase of the system clock. The limit of 500 ns may be exceeded as an accumulated error over
many PCR values. However, when the accumulated error spans a sufficiently long duration, it should be considered in
terms of its drift contribution and, allowed to exceed the 500 ns limit. What sufficiently long means has been derived in
clause 5 of this annex and is represented graphically by the break points of the graph I-2. For sinusoidal frequencies
lower than 12 mHz the limit is set by the drift rate specification rather than by the 500 ns limit.
ETSI
152 ETSI TR 101 290 V1.2.1 (2001-05)
If appropriate filters are built into the measurement device to separate the received PCR value spectral components
around a jitter vs. drift demarcation frequency, then it is possible to compare the errors received against the appropriate
limits set by the Standard.
Should the design of the measurement device be built as analogue device with hardware filters, then the designer will
use the demarcation frequency as a requirement for the design of the filters with independence of the sampling rate at
which the PCRs are actually arriving. This demarcation frequency is derived from the limits set in the Standard and
does not depend on sampling rate for the PCR values.
If the design of the filters is done by DSP techniques, the designer must take into account the average sampling rate of
the PCR values and adapt the filters to maintain a relatively fixed bandwidth for the measurement. This approach
implicitly assumes that the sampling rate (average arrival rate of PCR values) is not only known but is relatively
constant.
A good recommendation is to have the value of the coefficients determined adaptively by measuring the actual arrival
rate of PCR values. In other words, use an adaptive filter with the variable parameter being the measured PCR rate.
This approach, has been tested in practice using very strong frequency modulation for the PCR values rate and the
results in the measured jitter and drift do have a very close correlation (within the accuracy limits of the measuring
device) to the jitter and drift errors inserted by the test generator into the PCR values under test. Generally, small
differences in measurement filter bandwidths do not affect jitter measurement results significantly since the jitter
spectral components are most often broad band. In fact, the order of the filter is most important since this determines the
filter output sensitivity to out-of-band components, which may have small amplitudes but very high first and second
time-derivatives.
Another consideration to have in account is not related to the verification of stream validity but is related to a debugging
tool to find the origin of the jitter should it exist and have certain periodicity or resonant frequencies. This tool is to
apply Fourier analysis to the received sampled data.
Again, for this type of analysis to be valid, it is assumed that the sampling rate is known and is regular. Then the
sampling rate has to be measured in order to know frequencies analysed in each frequency bin (the resolution as a
function of the number of time domain samples used in the calculation and the relative stability of the sampling rate
over the measurement interval).
The problem of the non-uniformity of the sampling rate could be overcome by careful interpolation before the Fourier
technique is applied. In general this interpolation is not necessary due to the fact that as a debugging tool, the need is
not to know what is the “exact" value of the frequencies and its amplitudes. What is needed is only to obtain an idea on
whether the jitter is just random or it has some predominant frequencies embedded.
Generally, when a Fourier analysis is done on regularly sampled signals and there is a stable sinusoidal component on
the signal, it's parameters can be obtained with great accuracy and a clear spectral line could be displayed with such data
represented as in a spectrum analyser. If the sinusoidal component were not stable then a broad spectral line with
lowered amplitude would be expected, broader and lower as greater is the FMimplicit in such a sinusoid.
If a stable sinusoid is present but the sampling rate is FMmodulated, as is the case of PCR arrival rate, then a broad and
lower spectral line can be expected, just similar to the previous case described. When a great deal of FM (random or
not) is present in the sampling signal, the spectrum becomes broader with less amplitude in each bin. . However as a
diagnostic tool it may still be valid.
I.9 Choice of filters in PCR measurement
I.9.1 Why is there a choice ?
PCR measurement is a difficult task. The PCR values do not occur very often and when they do, they are rather large
(42 bit) numbers. The Clock reference is intended to be very stable, and as such a measurement device must have at
least the same stability to make a measurement. It is this long term stability (of the order of a few ppm change in
frequency per hour) in a counter which is incrementing very fast (27 MHz), but transmitted infrequently (40 ms or so)
which causes the problems.
ETSI
153 ETSI TR 101 290 V1.2.1 (2001-05)
A "Demarcation" frequency has been defined (figure I-2) which is able to divide the innacuracies added to the PCR
clock into Drift (low frequency component) and Jitter (high frequency component). It is based on the limits indicated in
ISO/ IEC 13818-1 [1] that sets a region below 10 mHz (MGF1) where the drift limit (75 mHz/s) is dominant and a
region above 10 mHz (MGF1) where errors are allow to exceed the drift rate but not the phase error limit (500 ns) that
is whyMGF1 is the highly recommended demarcation frequency used for accurate compliance to ISO/IEC 13818-1 [1].
For practical measurements, however, three fixed demarcation frequencies have been specified MGF1-3 and a user or
manufacturer defined one is also allowed MGF4. The demarcation frequency chosen is a compromise between the
desired accuracy of the clock as defined in the MPEG specification, and the practical concerns with performing the
measurement..
In order for two measurement devices to give the same results for a given transport stream, they must use the same
demarcation frequency in the measurement. In addition, any secondary effects due to irregular arrival of the PCR
samples may be removed so that results may match more closely. The way this is done is beyond the scope of this
measurement guideline, but designs should give similar results when, say, a 10 minute stream has PCRs every 20ms for
the first 5 minutes and then 40 ms for the next 5 minutes.
When the filter profiles MGF1 to MGF4 defined in the present document are implemented, there will be deviations
between the real response of the filters and the desired response of the ideal filters. This will give some measurement
errors between devices. In general, the precision of the filtering is a commercial choice of the equipment manufacturer
who is building equipment for a specific market.
The choice
The guidelines, PCR reference model and bitstream model are all intended to create an environment where similar
machines give similar results, and users are able to understand the implications of choosing different measurement
parameters. The errors between different devices will vary depending on a number of factors:
1) Are the same demarcation frequencies being used? This is the major factor.
If different devices use different demarcation frequencies then they will give different results. This will be a
major source of error. A discussion of the nature of the error is given below.
2) Are the demarcation filters of the same order? This is less important
If one device uses a 2nd order filter and another uses a 5th order filter then the nature of the filter response will be
quite different. There is likely to be a small difference between measurement devices particularly if significant
frequency components of the errors are close to the chosen demarcation frequency.
3) Is the measurement being made near the crossover of the offset/drift/jitter frequencies?
Near the crossover frequency, the order of the filter and its impulse response are likely to affect the frequency
components which are included or rejected from the measurements. This has much less of an affect than the
choice of demarcation frequency.
I.9.2 Higher demarcation frequencies
There are several effects of choosing a higher demarcation frequency (e.g. MGF3):
1) Jitter turns into drift or frequency offset.
A higher demarcation frequency means that frequency component which would have been classed as jitter
will now be classed as frequency offset or drift. This has the effect of reducing the magnitude of the overall
jitter frequency component. It also makes the system clock look less stable than it actually is.
2) The measurement settles faster.
The settling time is closely related to 1/frequency. If the frequency is increased by two orders of magnitude,
then the settling time may be reduced by two orders of magnitude. There are DSP techniques which can be
used to improve settling times, and the use of these is a commercial choice of the equipment vendor.
As a rough rule of thumb: a higher demarcation frequency settles faster but gives a less accurate result. Jitter
measurements should appear smaller and drift measurements should appear larger.
ETSI
154 ETSI TR 101 290 V1.2.1 (2001-05)
I.9.3 Lower demarcation frequencies
There are several effects of choosing a lower demarcation frequency (e.g. MGF1):
1. separation of drift and jitter into more representative groupings.
A lower demarcation frequency means that frequency components are more accurately classed as jitter,
frequency offset or drift. This has the effect of measuring the frequency components based on assumptions
which are closer to the values in the MPEG2 specification.
2. The measurement takes longer to settle.
The settling time is closely related to 1/frequency. If the frequency is reduced by two orders of magnitude,
then the settling time may increase by two orders of magnitude. There are DSP techniques which can be used
to improve settling times, and the use of these is a commercial choice of the equipment vendor.
As a rough rule of thumb: a lower demarcation frequency settles more slowly but gives a more accurate result. Jitter
measurements should appear larger and drift measurements should appear smaller.
The final choice of demarcation frequency rests with the user of the equipment and will come down to a trade off
between speed of measurement and precision of measurement. These guidelines should allow different measurement
devices to give comparable results in the heart of the measurement region, some ambiguity at the crossover point and
then agreement in the next region.
I.10 Excitation model for PCR measurement devices
I.10.1 Introduction
Work has been ongoing to define PCR measurements such that different equipment may show identical PCR measures
when given the same stimulus. Extensive work has been carried out on defining the demarcation frequencies and
relationships between parameters. In particular, practical definitions of the limits on timing error, d.c. offset and drift
can now be created with reference to the MPEG values set in ISO/IEC 13818-1 [1] .
In order to correctly test a system, however, a known good stimulus is required. This informative annex defines an
excitation model for PCR measurements which could be applied to an on-line or off-line system to ensure that the
measured PCR parameters arose as a result of the system, rather than a faulty source. In addition, a set of filters for
analysing PCRs could be tested so that, regardless of implementation, consistent results would be given for an identical
input. This annex is intended to outline the protocol for MGF1 testing for both network and device excitation.
ETSI
155 ETSI TR 101 290 V1.2.1 (2001-05)
Outline of the method
A multi-program, multi-PCR transport stream can be defined which can be used as a conformance stream for the
measurement device. The stream would have the following properties:
Component Description of measurement results
Service 1 Perfect PCR with regular intervals between samples
f PCR (t) = f o
Service 2 Perfect PCR with irregular intervals between samples
f PCR (t) = f o
Service 3 Frequency offset only
f PCR (t) = f o + f dc
measured PCR dc drift jitter
meas
f PCR (t) = f (t) ± e ± e ± e
Service 4 PCR drift and (unavoidable) jitter
f PCR (t) = f o + Am fm cos(2πfmt)
measured PCR dc drift jitter
meas
f PCR (t) = f (t) ± e ± e ± e
Service 5 PCR jitter only
f PCR (t) = fo + f j (t)
measured PCR dc drift jitter
meas
f PCR (t) = f (t) ± e ± e ± e
where f PCR (t)means instantaneous frequency, fo = 27,000 000 MHz , fdc is the offset frequency, fm is the drift
frequency, and f j (t) represents the instantaneous frequency of a jitter source. The values edc , edrift and e jitter
are error ranges which may be different for MGF1, MGF2 and MGF3 criteria.
This transport stream is defined in a pseudo-code so that it can be simply and unambiguously synthesized on a
computer. It would be appropriate for off-line testing as well as on-line playback from a suitable player. The stream
would have enough PSI to bind the stream, but SI or other components are outside the scope of the present document.
The stream may be constructed in such a way as to show independence between measurement accuracy and irregular
arrival of PCR values.
I.10.2 Constraints on the definition of a stream
This excitation model defines a stream which may be used both online and offline. In order to be used online, a
"perfect" bitstream player is required. This topic is outside the scope of the present document, but for now, let's assume
such a thing exists.
1) In many practical situations, a test transport stream needs to be generated at a specific bitrate (e.g. for a UK
DVB-T emission, a stream of 24,128 342 MHz might be desirable).
2) To comply with DVB guidelines, it is often desirable to fix the PCR insertion rate at some value less than 40 ms
in accordance with TR 101 154 [4].
3) The reference PCR in the excitation model should appear perfect. In order to achieve this, the sampling point of
the time reference (see note) should appear to be on a 27,0000 MHz sampling grid, and simultaneously on a
188 byte packet grid. i.e. each PCR sample is exact and has no quantization errors.
NOTE: ISO/IEC 13818-1 [1] clause 2.4.3.5 definition of program_clock_reference_base states the PCR is valid
on receipt of the last byte of program_clock_reference_base.
4) The insertion rate of the PCRs should meet the desired tolerances. A PCR measurement device should give
identical results, regardless of the insertion rate of the PCR samples.
5) A variable insertion rate may be one cause of measurement inaccuracy. The simplest of the perfect PCR services
should therefore have strictly regular PCR insertion rate, with a second perfect PCR service carrying pure values
but on an irregular grid.
ETSI
156 ETSI TR 101 290 V1.2.1 (2001-05)
Requirements 3 and 5 are hard requirements which must be satisfied to create a perfect stream. The other requirements
have some flexibility which allows us to create useable streams.
It is then possible to create a multi program Transport Stream with a perfect PCR and a perfect frequency offset if the
overall bitrate of the stream is carefully chosen. However, perfect drift ( edrift = 0 ) is not attainable in practice because
of quantization errors which also introduces an unavoidable high frequency jitter ( e jitter ) component. Nonetheless, it is
possible to reduce this noise to some degree by careful choice of the sampling points.
In general, the addition of jitter must be done in a band limited way to prevent aliased components of the jitter being
mirrored back into the frequency bands for drift and offset measurement. This is not representative of true jitter, but is
essential for this model which is intended to verify the implementation of a set of filters which meet the conditions for
the profiles proposed (see Break frequencies in clause I.5. In addition, this creates a useful stimulus for verifying/testing
jitter correction devices in network scenarios.
Definitions
Although one would ideally like to use the bitrate as the control parameter, the condition that the PCR samples fall on
both a 27 MHz grid and a 188 byte packet grid means that it is more practical to define a minimum time interval
between PCRs (which falls on the 27 MHz grid) and then set the bitrate by defining how many (whole) 188 byte
packets we wish there to be in this interval. In other words, defining the period of the beat frequency between 27MHz
and the packet rate. This effectively quantizes the values of achievable bitrate. It does not necessarily mean that PCRs
will appear in the stream with this minimum 'beat interval' separation - it just sets the granularity of insertion points. If
we wish to be able to allocate irregular inter-PCR spacings over a range of say 5 ms to 40 ms, then it is futile to set the
beat interval somewhere in the region of, say, 38 ms, since the legal values of the interval would be multiples of 38ms,
e.g. 38 ms, 76 ms, 114 ms,… etc. What is desirable is a beat interval with relatively fine granularity, so that there are a
number of legal insertion points compliant with TR 101 154 [4]. The trade-off is that the shorter the beat interval, the
coarser the quantization of the allowed values of bitrate become.
The actual beat interval is related to the desired beat interval by:
s
27 000 000
n
Ta =
where:
n = int (Td ×27 000 000)
being the integer number of 27 MHz clock pulses between PCRs. The range of possible bitrates that can be achieved
with this actual minimum time interval is:
bit/s
188 8




× = ×
a
a T
B p
where p is an integer. The values of p and Td can now be found which reduce the beat interval error and the bitrate
error (relative to the desired bitrate, Bd ) defined as:
106 ppm
int ×
−
− =
d
a d
beat T
T T
err and ×106 ppm
−
=
d
a d
bitrate B
B B
err
The values p and Td are the master values used to govern the creation of the excitation test stream. For regularly
spaced PCR samples we similarly define the actual regular spacing Ra in terms of the desired regular spacing, Rd , as,






= ×
a
d
a a T
R
R T int
with the corresponding PCR interval error,
ETSI
157 ETSI TR 101 290 V1.2.1 (2001-05)
106 ppm
int ×
−
− =
d
a d
PCR R
R R
err
The number of packets between the regularly spaced PCR samples is simply:
188×8
×
= a a B R
P
which is, by definition, an integer. If the desired length of the stream is defined as an integer number, F of 25Hz
frames, then the desired duration in seconds, Ld is just F / 25 s. The closest achievable length in units of 188 byte
packets is:
2
1
188 8
int +
×
= a × d
L
B L
P
d
And the closest achievable length in units of P packets (i.e. an integer number of regularly spaced PCR samples) is:






= +
2
1
int
P
P
P d
a
L
R
The achievable stream length is then:
a
R
a B
P P
L a
× ×188×8
=
The stream length error between the desired and achievable is then,
= − ×106 ppm
d
a d
length L
L L
err
The length of the stream should exceed the settling time of the measurement filters. This is difficult to define rigorously,
but must certainly exceed the drift/jitter demarcation frequency period of
11,86mHz
1
84,3 s = (see clause I.5). The
detection of drift frequencies in the region of say 1 mHz requires significantly longer than this.
To create the services it is possible to use a mathematical model to derive the clock pulse count N(t ) as a function of
time and use this count to create PCR values according to the definition of PCR i.e. including wraparound.
Service 1 (perfect service with regular inter-PCR spacing)
This is the simplest of the services. The clock count used to stamp the PCRs can be modelled by
N(t ) = f pt
If the inter-PCR timing is chosen to be i × n where i is an integer (and n is defined above as
n = int (Td ×27 000 000)), then the clock pulse count for the m th PCR sample is
N(mTa ) = m×i×n
subject of course to the constraint on maximum inter-PCR spacing, i ×n×Ta ≤ 40ms
Service 2 (perfect service with irregular inter-PCR spacing)
Similarly to the above, the clock count used to stamp the PCRs is still:
N(t ) = f pt
ETSI
158 ETSI TR 101 290 V1.2.1 (2001-05)
However, every sample in the stream is separated from the previous sample by a random integer multiple of n clock
pulses rather than exactly i × n . This is again subject to the constraint on maximum inter-PCR spacing, so the maximum
allowed multiple is:






×
× −
n Ta
40 10 3
int
Service 3 (pure offset)
For this service, the clock count used to stamp the PCRs is modelled by
N(t) = (f p + fdc )t
In order to eliminate any quantization errors (and hence jitter) from this service, we must choose the offset
frequency fdc so that:
fdcTa = j = integer valued
The offset means that against our 27 MHz timing grid, the clock used to stamp the PCRs is running either faster or
slower according to the sign of fdc . Similarly to the above we choose to space samples irregularly so that every sample
in the stream is separated from the previous sample by a random integer multiple of n + j clock pulses.
Service 4 (drift service)
For this service, the drift is modelled by a harmonic modulation so that the clock count used to stamp the PCRs is:
( ) ( f t)
A
N t f t m
m
p π
π
sin 2
2
= + (equation 1)
Within the constraints of the DVB recommendations on maximum inter-PCR times, it is impossible to create a stream
that contains a legal drift component without quantization errors (and hence jitter). This unavoidable source of jitter
introduces a maximum absolute timing error of one clock pulse. Although this cannot be eliminated, we can attempt to
minimize it by 'cherry picking' the PCR insertion points to reduce the error as much as possible. As with the first two
cases above, the fundamental unit of time between PCR samples is represented by n clock pulses. For each new
sample, all possible choices of time increment in the range; nTa , 2nTa , 3nTa , , mrangenTa K are considered and one
with the minimum the quantization error is chosen. The upper end of the range is bounded by mrangenTa ≤ 40 ms.
Service 5 (pure jitter service)
The creation of pure jitter is non-trivial. The clock count used to stamp the PCRs is defined by
N(t) = f pt + J (t )
Where J(t ) is a jitter source which models clock/network jitter in such a way that the resulting PCRs exhibit no d.c.
offset, or any fluctuations in the drift region of the spectrum. MGF1 defines the demarcation frequency between drift
and jitter as 10 mHz. Therefore J(t ) should not introduce any significant fluctuations below 10 mHz. In practice, there
is no upper bound on jitter timing error and unfortunately the relatively low sampling rate for PCR insertion inevitably
leads to aliasing of high frequency jitter. For the purposes of test, we choose to define our model jitter source J(t ) in
such a way that we avoid this aliasing. PCR samples separated by 100 ms - the maximum allowed interval under the
MPEG specification – have a corresponding Nyquist frequency of 5 Hz. Clearly therefore, our jitter source must not
contain any significant fluctuations above 5 Hz. These two frequencies set bounds on the spectral components
permissible in J(t ). In addition, the jitter source should be designed such that the maximum absolute clock error is as
close as possible to the MPEG limit of ±500 ns.
I.10.3 The Algorithm
There are 3 stages to the algorithm: Parameterization, Scheduling and Synthesis.
ETSI
159 ETSI TR 101 290 V1.2.1 (2001-05)
I.10.3.1 Parameterization
This is the first stage. This involves selecting the parameter values used to make the stream. These are the values for
Td and P that minimize the bitrate error and insertion rate error, and specifying the duration of the transport stream. It
also involves making a choice for the d.c. offset, f dc , and the drift frequency fm . The choice of fm determines the
drift amplitude, Am for maximum drift since, by definition,
maximumdrift 75 mHzs-1 2 2 = = πAm fm
In addition to this, there is the constraint that the frequency offset must not exceed ±800 Hz which means that:
810
2
≤
π
Am
I.10.3.2 Scheduling
This is the second process carried out. Each packet to be created is assigned a PID value so that the error involved in
creating the PCRs for each service is minimized. This process is performed on a component by component basis until
all the criteria have been satisfied. The regularly sampled perfect service is inserted first, taking the packets required for
regular spacing. The d.c. offset service is inserted next, using packet choices that do not clash with the first service. The
drift service follows, using unallocated packets that minimize the quantization error. The irregularly sampled perfect
service and jitter service are inserted last since these have the greatest degree of flexibility over where their packets lie.
I.10.3.3 Synthesis
Finally, the pre-allocated packet structure is synthesized into a valid transport stream. The multiplexing of valid video
and audio content is outside the scope of the present document. Only empty packets will be covered in the pseudo code
given here.
I.10.4 The Pseudo-C code
The excitation model is written in Pseudo-C and can be used to generate a file where the 1st service will have a perfect
PCR.
/* All values are defined and fixed and should not be changed
Time is tracked by a 27MHz pulse count index which is passed to the subroutines
The bitrate and other values have been adjusted to work.
Rand() is a function that returns a uniform deviate in the range 0 to 1.
original: BFD 27 Nov 1999
r1: BFD 25 Jan 2000
r2: BFD 20 Feb 2000
r3: JD 2 May 2000
*/
/**************************************************************************************/
/* Parameters for the model */
/**************************************************************************************/
#define PATsPerSecond 20
#define PMTsPerSecond 20
/* ------- define constants and fixed values ------- */
#define Pi 3.1415926535897932384626433
#define SCR 27000000 /* System Clock Frequency in Hz */
#define PCRDriftRate 0.075 /* maximum drift rate in Hz/second */
#define PCRMaxSpacing 40e-03 /* maximum desired inter-PCR spacing in second */
/* ------user-defined parameters (below is simple stream example from appendix A)-----*/
#define n 172800 /* user defined inter-PCR minimum # 27 MHz clock pulses */
#define i 5 /* user defined # of n's between regular PCR samples */
#define Ta 0.0064 /* user determined ACTUAL min inter-PCR timing in seconds*/
#define Fdc 781.25 /* user defined offset value in Hz */
#define La 240/* user defined length of stream in seconds */
#define Fm 0.005 /* user defined drift frequency in Hz */
ETSI
160 ETSI TR 101 290 V1.2.1 (2001-05)
/* ------- dependent parameters ------- */
#define Total_count(SCR*La) /* # 27MHz clock pulses in entire stream */
#define Am (PCRDriftRate /(2.0*Pi*Fm*Fm)) /* dimensionless drift amplitude */
#define N (n*i) /* #clock pulses between regular PCRs */
#define mrange (PCRMaxSpacing/(n*Ta)) /* max # of n's between two PCRs */
#define J (Fdc*Ta)
#define N_off (n+J) /*min clock pulses between offset PCRs */
#define N_offrange (PCRMaxSpacing/(N_off*Ta))/* max # of (n+J)s between offset PCRs */
/**************************************************************************************/
/* Data creation */
/*****************/
/*
Create the PID array.
*/
/**************************************************************************************/
CreatePIDArrays()
{
/* Using an appropriate storage mechanism */
/* must store: PCR value & PID of each packet */
}
/* Insert Perfect Packets (on regular grid) according to embedded algorithm */
Schedule_RegularPerfectPCRPackets()
{
clock_count =0;
while(clock_count<Total_count)
{
clock_count += N;
RegPerfectPCR = PCR(clock_count);
AllocatePacket(clock_count, RegPerfectPCR, RegularPIDvalue);
}
}
/* Insert Perfect Packets (on irregular grid) according to embedded algorithm */
Schedule_IrregularPerfectPCRPackets()
{
clock_count = 0;
while(clock_count<Total_count)
{
Successful = FALSE;
while(!Successful)
{
trial_clock_count = clock_count + n*(int)(mrange*Rand());
IrregPerfectPCR = PCR(trial_clock_count);
Successful = AllocatePacket(trial_clock_count, IrregPerfectPCR
, IrregularPIDvalue);
}
clock_count = trial_clock_count;
}
}
/* Insert Drift Packets according to embedded algorithm */
Schedule_DriftPackets()
{
clock_count = 0;
while (clock_count<Total_count)
{
MinQE = 1e30;
best_m = 1;
trial_fp_clock_count = (float) clock_count;
/* check all possible available packets & choose one with least quantisation error */
for(m=1, m<mrange+1; m++)
{
clock_increment = n*m;
trial_fp_clock_count += clock_increment;
model_time = trial_fp_clock_count/SCR;
trial_fp_clock_count += (Am/(2.0*Pi))*sin(2.0*Pi*Fm*(model_time));
/* ref eqn 1 */
DriftPCR = PCR(trial_fp_clock_count);
vacant = Check_PID_Vacancy(clock_count + clock_increment);
if(vacant)
{
QE = AbsQuantizationError(trial_fp_clock_count, DriftPCR);
if(QE<MinQE) /* keep track of packet with least
quantisation error */
ETSI
161 ETSI TR 101 290 V1.2.1 (2001-05)
{
MinQE=QE;
best_DriftPCR = DriftPCR;
best_m=m;
}
}
}
clock_count += n*best_m;
DriftPCR = best_DriftPCR;
AllocatePacket(clock_count, DriftPCR, DriftPIDvalue);
}
}
/* Insert Offset Packets according to embedded algorithm */
Schedule_OffsetPackets()
{
clock_count = 0;
while (clock_count<Total_count)
{
Successful = FALSE;
while(!Successful)
{
trial_clock_count = clock_count + n_off*(int)(n_offrange*Rand());
OffsetPCR = PCR(trial_clock_count);
Successful = AllocatePacket(trial_clock_count, OffsetPCR
, OffsetPIDvalue);
}
clock_count = trial_clock_count;
}
}
/* Insert Jitter Packets according to embedded algorithm */
Schedule_JitterPackets()
{
clock_count = 0;
while (clock_count<Total_count)
{
Successful = FALSE;
while(!Successful)
{
trial_clock_count = clock_count + n*(int)(mrange*Rand());
trial_fp_clock_count = trial_clock_count + JitterSource();
JitterPCR = PCR(trial_fp_clock_count);
Successful = AllocatePacket(trial_clock_count, JitterPCR
, JitterPIDvalue);
}
clock_count = trial_clock_count;
}
}
/* Insert PATs as required */
Schedule_PATPackets()
{}
/* Insert PMTs as required */
Schedule_PMTPackets()
{}
/* Insert Null packets as required */
Schedule_NullPackets()
{
}
JitterSource() //band limited jitter source
{}
PCR(clock_count) //PCR values made using the extension/base convention with wraparound
{}
Check_PID_Vacancy(clock_count)
{
ETSI
162 ETSI TR 101 290 V1.2.1 (2001-05)
}
AllocatePacket(clock_count, trialPCR, PIDvalue)
{
if(Check_PID_Vacancy(clock_count))
{
ReservePacket(clock_count, trialPCR);
return TRUE;
}
else
return FALSE;
}
main()
{
/* The first step is to create a large empty array */
CreatePIDArrays();
/*
Now we schedule all the packets of the different services to ensure
that we create a stream with the lowest quantisation errors
*/
Schedule_RegularPerfectPCRPackets();
Schedule_OffsetPackets();
Schedule_DriftPackets();
Schedule_IrregularPerfectPCRPackets();
Schedule_JitterPackets();
/*
Now insert the PSI to bind the stream together
*/
Schedule_PATPackets();
Schedule_PMTPackets();
Schedule_NullPackets();
/*
Finally it is time to synthesise the final data
*/
SynthesiseTS("PCRverify.m2t");
}
I.10.5 Parameter definitions and example values
The following table lists some example values of the user defined parameters where 'PCR spacing' refers to the spacing
of regularly sampled 'perfect' PCRs. The parameters in bold are the independent ones used in the model. The quantities
within the outlined boxes are the desired parameter values.
ETSI
163 ETSI TR 101 290 V1.2.1 (2001-05)
Parameter Description Simple stream DVB-T like DVB-S like
Td Desired beat spacing in ms 6,4 10,036 10,009 65
Ta Achievable beat spacing in ms 6,4 10,036 10,009 629 63
n 27 MHz pulses between beats 172 800 270 972 270 260
errbeat−int Beat interval error in ppm 0,00 0,00 2,04
Bd Desired bitrate in bit/s 470 000 24 128 342,00 380 147 06
Ba Achievable bitrate in bit/s 470 000 24 127 540,85 38 014 593,35
p Packets between beats 2 161 253
errbitrate Bitrate error in ppm 0,00 33,20 2,96
Rd Desired inter-PCR spacing in ms 32 30,108 30,029
Ra Achievable inter-PCR spacing in
ms
32 30,108 30,028 888 89
errPCR−int PCR interval error in ppm 0,00 0,00 3,70
P Packets between PCRs 10 483 759
F Desired length in 25 Hz frames 6 000 5 250 3 390
Ld Desired length in seconds 240 210 135,6
Ld P Closest integer # packets to Ld 75 000 3 368 872 3 427 380
Ra P Total # packets in stream when
Ld P is quantized to P
7 500 6 975 4 516
La Duration of
Rd P packets in
seconds
240 210,003 3 135,610 462 2
errlength Length error in ppm 0,00 15,71 77,16
Fs Stream size in MBytes 14,1 633,357 9 644,397 072
d
dc F Desired d.c. offset frequency in Hz 810 810 810
a
dc F Nearest attainable frequency to
d
dc F in Hz
781,25 797,130 330 8 799,230 370 8
a
j = Ta Fdc 5 8 8
fm Drift modulation frequency in Hz 0,005 0,005 0,005
Am Drift modulation amplitude 477,464 829 28 477,464 829 28 477,464 829 28
2 2 πAm fm Maximum absolute drift in mHz/s 75 75 75
2π
Am Maximum drift frequency
excursion in Hz
75,990 887 73 75,990 887 73 75,990 887 73
a
a
T
R
i =
Number of beat intervals between
regular PCR samples
5 3 3
ETSI
164 ETSI TR 101 290 V1.2.1 (2001-05)
Annex J (informative):
Bitrate related measurements
J.1 Introduction
J.1.1 Purpose of bitrate measurement
This annex is intended to clarify a bitrate measurement technique which will allow different vendors of equipment to
display the same bitrate value on their equipment when they analyse the same transport stream.
The measurement technique in this specification should be applicable to the whole transport stream as well as its
individual components. This should allow displays of transport stream information such as the traditional "bouncing
bars" statistical multiplex display to be shown consistently on different equipment. This display is intended to
dynamically show the different allocation of bitrate between different services. The intention is that the measurement
should be stand-alone and non-intrusive.
The measurement technique should also be easy to implement so that cost-effective designs can be introduced to large
MPTS systems. It should also be scalable so that as extra precision is required, a more expensive device can be built
using the same principles.
The technique is also appropriate for non Transport Stream system, but the use in such systems is outside the scope of
the present document.
J.1.2 User Rate versus Multiplex Rate
MPEG-2 transport streams are comprised of many different elements including but not limited to multiple compressed
video and audio streams, teletext, table data, conditional access streams, IP data, and other private data. Each of these
individual elements and the overall transport stream have data rates associated with them. The data rates can be time
varying for the individual elements and the overall stream.
It is of importance to define the measurement of these rates and have a common definition for these measurements.
Before the measurements can be defined, the multiplexing of all the elements into a transport stream needs to be
understood with regards to rate calculations.
Figure J.1 depicts a general representation of the multiplexing process.
ETSI
165 ETSI TR 101 290 V1.2.1 (2001-05)
Video 1
Audio 1
Video 2
...
Audio 2
PSI/SI
Tables
Null PID
Multiplex
Switch
User
Rates
Multiplex
Rates
Buffers
Transport
Stream
Packets
...
...
Rate
Measurement
Device
Figure J-1: General representation of the multiplexing process
This diagram represents a number of different elements being multiplexed into a single transport stream. Before all the
streams are multiplexed together they can be considered to have User rates which are established by the user (e.g.
4 Mbits/s for Video 1). It can be modeled that each element has a User data rate entering the buffer and a Multiplex rate
leaving the buffer since the data is extracted directly from the buffer and placed as a complete packet in the transport
stream. Over the long term average, the User and Multiplex rates must be the same, but the creation of the transport
stream through the multiplex process can either increase or decrease the User rate in the actual transport stream over a
specific Time Gate. For example, the video might have a 4,1 Mbits of data over a one-second Time Gate in the transport
stream, but in the next one second interval it could have 3,9 Mbits. But with respect to the PTS/DTS values in the
stream, the video rate as set by the user could still be 4,0 Mbits/s.
The Multiplex rates will also depend upon what is actually being multiplexed together, and the measurement of the
multiplex rate in the output stream will vary if different elements are combined. If only one video is being transmitted at
one time and another video is being transmitted at another time, the output Multiplex rate will be different at those two
times even if the User rate has not changed.
The User rate for video also needs to be better understood since a single number is often given for this rate (e.g.
4 Mbits/s). This number typically means the total number of bits in a GOP multiplied by the number of GOPs per
second. The actual rate of video varies with each frame. An I frame typically receives a much higher percentage of the
bits compared to the B and P frames. What generally happens is that even though the I frame has significantly more
data than a B frame, it will take longer to transmit this frame and the Multiplex rate can approach the User rate. This
definition of User rate for video applies to both the CBR and VBR approaches. In the CBR case, the user provides one
value for the rate, while in the VBR case the user provides a minimum and maximum and typically lets compression
equipment vary the rate between these parameters in order to maximize video quality based on some constraints. The
rate as calculated by the compression equipment is still considered a User rate since it is before the video data is
multiplexed into the transport stream.
Since the rates of the elements are less than or equal to the rate of the output transport stream, the positioning of these
elements in the output stream is important to consider in calculating the User rate. For example, an element that
generates 10 packets per second may have these packets placed at the beginning of the second, in the middle, dispersed
throughout, etc. Buffer models in general restrict the packet placement but as an extreme example, it could be assumed
that the packets are placed at the beginning of a second and the transport rate is 1,5040 Mbits/s. If the Time Gate of a
rate measurement of this element is 0,1 s and this Time Gate started with the transmission of these packets, the first rate
measurement would be 0,1504 Mbits/s. If the next measurement also uses 0,1 s of duration and starts just after the
packet is transmitted, the rate would be 0,0 Mbits/s. Neither of these numbers matches the expected User rate of
0,01504 Mbits/s.
ETSI
166 ETSI TR 101 290 V1.2.1 (2001-05)
A real world example for a 256 kbit/s audio stream can easily indicate differences of 2 % in the User rate versus the
Multiplex rate. This audio stream has approximately 200 packets per second with each audio frame containing about
5 packets. In a measurement interval of one second that begins in the second half of an audio frame, all 5 of the first
packets can be transmitted in the second half of an audio frame, and all 5 of the last five packets can be transmitted in
the first half of the last audio frame. These results in a Multiplex rate of 205 packets per second that is 2,5 % higher
than the User rate of 200 packets per second. This error difference can increase with smaller measurement intervals
since for a 100 ms interval the number of packets for the User rate would be 20 while the Multiplex rate could be 25
resulting in a 25 % difference.
J.1.3 User rate applications
The rate measurements for transport streams are computed for a variety of purposes. These include but are not limited
to:
- Verification/conformance/troubleshooting - the overall transport stream rate or rates of individual elements are
expected to be certain values as set by a user or compression/multiplex system. The user needs to validate that
the rates in the stream meet the "expected" rates. This validation can be done over time or just once and can
include statistics (e.g. minimum and maximum) as well as history of any rate calculation values. The validation
would include all elements including video, audio, conditional access data, PSI/SI tables, etc.
- Video and audio quality - there is a strong correlation between video and audio quality and the rate at which
these items are transmitted in the transport stream. There is especially a need to monitor the rate of the video
since this rate often varies over time and if an video quality issue is determined by visual inspection, there would
be a need to determine the rate of the video at that time. A service provider may also guarantee a minimum bit
rate for video and audio for a particular program and with a contract, this provider will need to prove that those
rates have been met.
- Sale of bandwidth - there is a need to monitor the rate of individual elements in a stream over a longer period so
that a service provider can charge a user for the bandwidth that has been used in one hour or one day or one
week, etc.
- Monitoring - there is a need to generate an alarm if the rate of a particular element or the whole stream goes
outside some user-specified minimum and maximum range. This error could mean that an element is no longer
being included in the transport stream due to a multiplexer malfunction. The accuracy of these rate
measurements is not critical to the overall application.
J.2 Principles of Bit rate measurement
This is a difficult subject as a measured bitrate depends on the time over which the bitrate is averaged. Bit rate is usually
expressed in terms of bits per second, but the actual value that is measured will depend on the way the bits are counted.
A bitrate measurement will depend on where in the system the bitrate is measured. For example, in a system, slightly
different bitrates may be seen depending on whether the bitrate is measured before or after a large buffer.
J.2.1 Gate or Window function
On the assumption that we are always dealing with Transport Stream packet based systems in the DVB world, we have
3 main choices when counting bytes:
- packet based - count only the synchronization bytes;
- byte based - count every byte when it arrives;
- bit based - count every bit as it arrives.
We also have 2 options for applying the window function:
- "continuously" rolling window;
- a jumping window (the end of each window is the start of the next window).
ETSI
167 ETSI TR 101 290 V1.2.1 (2001-05)
A jumping window is very undesirable as the bitrate measured will vary depending on when the window is first applied.
This rules it out very early. A rolling window is therefore more desirable, but some caution is needed in the use of the
term "continuous".
The most precise bitrates would be given with a bit based counting scheme. Here, each time a new bit is received, or
sent, the total number of bits in the last time window (e.g. 1 second) could be counted and a value displayed. This
would always give the most accurate value, but there are a number of serious technical difficulties in implementing this,
particularly in offline and semi-offline systems. These difficulties include processing bandwidth and timing accuracy. A
byte based system also requires large bandwidth, but both bit and byte based may be required in some special
circumstances. Although this specification does not to define byte or bit based profiles, they can easily be added by
counting the bytes or bits and adjusting the nomenclature appropriately.
A packet based approach may be favourable in situations where cheap implementations with reasonable accuracy are
required. It is likely that most DVB Tx and Rx systems would have the capability of deriving some timing information
on a packet basis.
J.2.2 "Continuous window"
If all transport streams were of a constant bitrate, not bursty, continuously clocked and could be easily analysed as a
signal with fixed and uniform temporal sampling, then bitrate measurement would be easy.
In real systems (bursty ASI, Transport streams over IP, 1394b hubs, cascaded networks etc.) the bytes and packets do
not necessarily arrive on a uniform sampling grid and pragmatic measures need to be taken in defining the window
function. To simplify implementation, we have looked at systems where the window function is moved across the data
in different ways: by byte, by packet, by fixed time interval.
There are several points to note about the algorithm in this specification:
1) Strictly speaking, this measure is not continuous.
2) It is a discrete measure whose bitrate values are only valid on time slice boundaries.
3) It is easy to implement and gives a new TS bitrate value every τ (11,1 μs to 1 s).
4) It is applicable to partial transport streams where only a subset of PIDs are being inspected.
5) It can be extended to measure the bitrate of the payload of TS packets.
6) It is repeatable between equipment vendors because the time slice can be made sufficiently small to ensure
aliasing is not a problem e.g. when τ = 1/90 kHz
J.2.3 Time Gate values:
20 ms: gives the peak bitrate of a stream based on variable bitrate elements within it.
1 s: gives a longer term "smooth" average.
user: could be used for elements such as subtitles which may only be present from time to time and may
require windows of 1 minute or more.
J.2.4 Rate measurements in a transport stream
Only the Multiplex rates are available to be measured in the transport stream and not the original User rates. In general,
it is the User rates that are of interest as outputs of a measurement device with some exception regarding issues of
burstiness and buffer models.
Depending on the customer application, the parameters that should be used in the MG bitrate equation in clause 5.3.3.
will be different if the user wants to measure User rates or Multiplex rates as finding the best accuracy for the User rates
is different than finding the best accuracy for the Multiplex rates. The parameters also need to take into account tracking
the changes in the rate versus time. The parameters should in general be different for elements that differ either in type
or in rate in order to maintain accuracy.
ETSI
168 ETSI TR 101 290 V1.2.1 (2001-05)
Here are some general considerations for the parameters:
- For elements that have CBR, increasing T will push the measured Multiplex rate towards the User rate.
- For reasonable accuracy of the User rate, T must be large enough to include multiple elements of what is being
measured. For example, if the rate of a SDT is being measured, it should include at least 10 different arrivals of
the SDT.
Decreasing τ will cause the Multiplex rate to be more accurately tracked but will not increase the accuracy of
calculating User rates for CBR streams. For VBR streams, a smaller τ to within some limits will allow the changes to be
better averaged over time.
J.3 Use of the MG profiles
The profiles in clause 5.3.3.2 have been designed to have the properties described below.
J.3.1 MGB1 Profile - the backwards compatible profile
This is a backwards compatible profile where a 1 second jumping window is used to measure bitrate. In a rigidly CBR
system, this will give a good indication of the bitrate, but will give aliasing and inaccuracy if the bitrate being measured
is changing faster than every 1s. This makes it impractical for looking at VBR systems, or for looking at the bitrates of
VBR components (e.g. stat-mux video) in a CBR transport stream.
This profile is included for backwards compatibility with existing equipment.
J.3.2 MGB2 Profile - the Basic bitrate profile
This profile is recommended for new designs. It is intended to give a good idea of the average bitrate of a system, yet
have enough resolution (due to a small τ value) to show whether the bitrate is truly static or is varying with time. The
values have been chosen to allow simple implementation.
J.3.3 MGB3 Profile - the precise Peak bitrate profile
This profile has a time gate which is small enough to show the variable bitrate characteristics of a statistical multiplex
environment. The timeSlice is small enough to ensure that only a single packet header will occur in each timeSlice for
most distribution systems. The time gate is short enough so that frame by frame averaging does not take place. The
timebase chosen can be locked to, or derived from the PCR in a decoder or encoder environment for ease of
implementation.
J.3.4 MGB4 Profile - the precise profile
This profile is intended to give a "true" smoothed bitrate. The timeSlice is small enough to ensure that only a single
packet header will occur in each timeSlice for most distribution systems. The time gate is a little over 1 second to give a
long time constant averaging to the data. The timebase chosen can be locked to, or derived from the PCR in a decoder
or encoder environment for ease of implementation.
J.3.5 MGB5 Profile - the user profile
This profile is intended to give extensibility to the bitrate measurement algorithm. It allows different time gates and
timeSlice values to be defined. These can be applied to the whole transport stream, or to individual components of the
stream. It is important when using this profile that the results are carefully documented using the nomenclature in these
guidelines. This will ensure that results can be repeated at a later date.
ETSI
169 ETSI TR 101 290 V1.2.1 (2001-05)
J.4 Error values in the measurements
It is worth noting the areas where errors can be introduced into the measurement:
• clock instability in the time gate and time slice functions;
• quantization due to counting elements which are too big e.g. too many or too few packet headers may fall within
the time gate;
• aliasing due to having a timeSlice or a time Gate which is too large for the parameter being measured.
In real systems, the errors due to clock instability and quantization tend to be rather small. The biggest problem is
inappropriate use of timeSlice and time gate values. This can be best demonstrated by an example.
Imagine a DVB-S statistical multiplex system (38.1 Mbit/s) where a particular video PID has a bitrate limit of
3 -5 Mbit/s and the hypothetical video encoder is able to change its bitrate every 80ms. Bit rate is measured by counting
packet headers of a certain PID. The average video rate is 4 Mbit/s.
If theMGB4 profile is used,
DVB-S ≈ 38,1 Mbit/s packet duration ≈ 40 μs packets per τ ≈ 0,25
The clock frequency error uncertainty may be as high as 500 ppm. This would lead to an error in the duration of the
time gate of 500 ppm (0,05 %). This could increase the 1 second window by 500 μs which at 5 Mbit/s could allow an
extra 2 packets into the gate. This would give an error of
=2 × 188 × 8 bits/s
=0,06 % of 5 Mbit/s
The uncertainty due to quantization is equal to the element size which is counted which is 1 packet per time gate in this
case
= 188 × 8 bit/s = 1 504 bit/s
= 0,03 % of 5 Mbit/s
It can be seen that these values are all quite small. If we imagine the slightly contrived example of a sequence which
requires the bitrate shown below:
Difficult
5 Mbit/s
1sec
Easy
3 Mbit/s
1sec
Difficult
5 Mbit/s
1sec
Easy
3 Mbit/s
1sec
Difficult
5 Mbit/s
1sec
Easy
3 Mbit/s
1sec
• The MGB4 profile will show a smoothed version of the above bitrate with peak values of 5 Mbit/s and 3 Mbit/s.
• The MGB3 profile will show much sharper edges to the bitrate changes and will report the peak values of
5 Mbit/s and 3 Mbit/s.
• The MGB1 profile, however will show different values depending on the moment when the 1 second window
jumps to its next starting point. If it is synchronized with the start of the 1 second sequences, then it will report
the correct values of 5 Mbit/s and 3 Mbit/s. If, however it starts its measurements 50 % of the way through a 1
second sequence, it will report that the bitrate is constant at 4 Mbit/s. This is an error of 33% at 3 Mbit/s or
20% at 5 Mbit/s.
Real errors are less than in this contrived example, but this source of error is the most significant in real systems. Note
that in some monitoring applications errors of a few percent may be tolerable, whereas in other applications a precision
of 1ppm or better may be required.
ETSI
170 ETSI TR 101 290 V1.2.1 (2001-05)
J.4.1 Very Precise measurements
In very accurate measurements, it may be necessary to count individual bytes, or individual bits to obtain the required
precision. The same algorithm, nomenclature and synchronization as described in clause 5.3.3 may still be used and the
results will be repeatable.
ETSI
171 ETSI TR 101 290 V1.2.1 (2001-05)
Annex K (informative):
DVB-T channel characteristics
This annex provides some information on terrestrial channel profiles which can be used for off-line computer
simulations and realtime simulations based on dedicated equipment. The properties of these profiles reflect realistic
reception conditions and/ or worst-case scenarios and were used to verify specific features of the DVB-T standard.
K.1 Theoretical channel profiles for simulations without
Doppler shift
(quoted from EN 300 744 [9])
The performance of the DVB-T system has been simulated during the development of the standard EN 300 744 [9] with
two channel models for fixed reception - F1 and portable reception - P1, respectively.
The channel models have been generated from the following equations where x(t) and y(t) are input and output signals
respectively:
a) Fixed reception F1:


=
=
⋅ + ⋅ − ⋅ ⋅ ⋅ −
=
N
i
i
N
i
i
j
x t i e x t
y t
i
0
2
1
2
0 ( ) ( )
( )
ρ
ρ ρ π θ τ
where:
- the first term before the sum represents the line of sight ray;
- N is the number of echoes equals to 20;
- θi is the phase shift from scattering of the i'th path - listed in table K.1;
- ρi is the attenuation of the i'th path - listed in table K.1;
- τi is the relative delay of the i'th path - listed in table K.1.
The Ricean factor K (the ratio of the power of the direct path (the line of sight ray) to the reflected paths) is given as:
=
= N
i 1
2
i
2
0
r
r
K
In the simulations a Ricean factor K = 10 dB has been used. In this case:

=
= ⋅
N
i
o i
1
ρ 10 ρ 2
ETSI
172 ETSI TR 101 290 V1.2.1 (2001-05)
b) Portable reception, Rayleigh fading (P1):
=
= ⋅ ⋅ − ⋅ ⋅ ⋅ −
N
i
i
j
y t k i e x t i
1
( ) ρ 2π θ ( τ ) where
=
=
N
i
i
k
1
2
1
ρ
θi, ρi and τi are given in table K.1.
Table K.1: Attenuation, phase and delay values for F1 and P1
i ρi τi [μs] θi [rad]
1 0,057 662 1,003 019 4,855 121
2 0,176 809 5,422 091 3,419 109
3 0,407 163 0,518 650 5,864 470
4 0,303 585 2,751 772 2,215 894
5 0,258 782 0,602 895 3,758 058
6 0,061 831 1,016 585 5,430 202
7 0,150 340 0,143 556 3,952 093
8 0,051 534 0,153 832 1,093 586
9 0,185 074 3,324 866 5,775 198
10 0,400 967 1,935 570 0,154 459
11 0,295 723 0,429 948 5,928 383
12 0,350 825 3,228 872 3,053 023
13 0,262 909 0,848 831 0,628 578
14 0,225 894 0,073 883 2,128 544
15 0,170 996 0,203 952 1,099 463
16 0,149 723 0,194 207 3,462 951
17 0,240 140 0,924 450 3,664 773
18 0,116 587 1,381 320 2,833 799
19 0,221 155 0,640 512 3,334 290
20 0,259 730 1,368 671 0,393 889
NOTE: Figures in italics are approximate values.
NOTE: For practical implementations profiles with reduced complexity have been used successfully. In many
cases it seems sufficient to use e. g. only the six paths with the highest amplitude.
K.2 Profiles for realtime simulations without Doppler shift
The following profiles were used in laboratory tests in a research project with satisfactory results.
NOTE: AC106 Validate (1995-1998).
Table K.2: Echo Profiles
Path fixed
delay [μs] C/I [dB]
Portable
delay [μs] C/I [dB]
dense SFN
delay [μs] C/I [dB]
#1 (main) 0 0 - - 0 0
#2 0,5 17,8 0,5 7,8 7,8 9,3
#3 1,95 17,9 1,95 7,9 11,6 5,5
#4 3,25 19,1 3,25 9,1 17,5 16,1
#5 2,75 20,4 2,75 10,4 20,0 14,5
#6 0,45 20,6 0,45 10,6 23,4 23,4
#7 - - 0,85 11,6 - -
ETSI
173 ETSI TR 101 290 V1.2.1 (2001-05)
K.3 Profiles for realtime simulation with Doppler shift
(mobile channel simulation)
In the course of a research project (see note), three channel profiles were selected to reproduce the DVB-T service
delivery situation in a mobile environment. Two of them reproduce the characteristics of the terrestrial channel
propagation with a single transmitter, the third one reproduces the situation coming from an SFN operation of the
DVB-T network.
NOTE: AC318 Motivate (1998-2000).
The following tables describe the composition of the chosen profiles.
• Typical Urban reception (TU6)
This profile reproduces the terrestrial propagation in an urban area. It was originally defined by COST207 as a
Typical Urban (TU6) profile and is made of 6 paths having wide dispersion in delay and relatively strong power.
This channel profile has also been used for GSM and DAB tests.
Tap number Delay (us) Power (dB) Doppler spectrum
1 0.0 -3 Rayleigh
2 0.2 0 Rayleigh
3 0.5 -2 Rayleigh
4 1.6 -6 Rayleigh
5 2.3 -8 Rayleigh
6 5.0 -10 Rayleigh
• Typical Rural Area reception (RA6)
This profile reproduces the terrestrial propagation in an rural area. It has been defined by COST207 as a Typical
Rural Area (RA6) profile and is made of 6 paths having relatively short delay and small power. This channel
profile has been used for GSM and DAB tests.
Tap number Delay (us) Power (dB) Doppler spectrum
1 0.0 0 Rice
2 0.1 -4 Rayleigh
3 0.2 -8 Rayleigh
4 0.3 -12 Rayleigh
5 0.4 -16 Rayleigh
6 0.5 -20 Rayleigh
• 0 dB Echo profile
This profile has been defined byMotivate partners. Its composition has been largely influenced by the specific
nature of the DVB-T signal, especially its spread spectrum technique (introducing an Inter Carrier Interference
sensitivity to Doppler spread) and its use of a Guard Interval (introducing an Inter Symbol sensitivity to the
echoes delays). Moreover, its definition has been driven by the analysis of the profiles encountered during the
various field trials performed during the Motivate project.
This profile is made of two rays having the same power, delayed by half the Guard Interval value and presenting
a pure Doppler characteristic.
Tap number Delay (us) Power (dB) Doppler spectrum Frequency ratio
1 0 0 Pure Doppler -1
2 1/2 Tg 0 Pure Doppler +1
ETSI
174 ETSI TR 101 290 V1.2.1 (2001-05)
Annex L (informative):
Bibliography
Proakis John G.: "Digital Communication", McGraw Hill, 1989.
Begin G., Haccoun D. and Chantal P.: "High-Rate Punctured Convolutional Codes for Viterbi and Sequential
Decoding", IEEE Trans. Commun., vol 37, pp. 1113-1125, Nov. 1989.
Begin G., Haccoun D. and Chantal P.: "Further Results on High-Rate Punctured Convolutional Codes for Viterbi and
Sequential Decoding", IEEE Trans. Commun., vol 38, pp. 1922-1928, 1990.
Odenwalder J.P.: "Error Control Coding Handbook", Final report prepared for United States Airforce under Contract
No. F44620-76-C-0056, 1976.
Pratt, Timothy and Bostian Charles W.: "Satellite Communications", John Wiley & Sons, 1986.
ETSI
175 ETSI TR 101 290 V1.2.1 (2001-05)
History
Document history
Edition 1 May 1997 Publication as ETR 290
V1.2.1 May 2001 Publication