Professional Documents
Culture Documents
Technical Guide
ANNEX B
OVERVIEW OF DVB
June 7, 1999
This document is provided for information purposes. Whilst every effort has been made to provide accurate information,
no responsibility is taken for errors or omissions. EUTELSAT reserves the right to change this information without notice.
Technical Guide
Annex B - Overview of DVB
CONTENTS
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1. Digital Versus Analogue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2. The DVB Standard(s) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3. DVB - A Mechanism for Data Broadcasting . . . . . . . . . . . . . . . . . . . . 2
1.4. Satellite Delivery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Basic Elements of A Satellite DVB System . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Baseband Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1. MPEG-2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2. Processing Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.3. Programmes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.4. The Transport Stream Multiplex . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5. Service Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.6. Conditional Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4. Satellite Channel Adaptation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.1. Service Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.2. System Description. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5. Contact Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1. Introduction
The DVB Project is run on a voluntary basis and brings together experts from
more than 200 companies and organisations, representing the interests of
manufacturing industry, broadcasters and services providers, network and
satellite operators and regulatory bodies. Its main intent is to reap the benefits
of technical standardisation, whilst at the same time satisfying the commercial
requirements of the project members. Although a large part of the
standardisation work is now complete, work is still ongoing on issues such as
the Multimedia Home Platform.
Much of the output of the DVB Project has been formalised by the European
Telecommunications Standards Institute (ETSI). A guide to the relevant ETSI
documentation is given in Annex B, together with references to other useful
documents.
1.5. Nomenclature
Certain differences exist in the terminology used within the MPEG, DVB and
traditional broadcasting communities. These differences are reviewed below
to facilitate understanding of the remainder of this annex.
In MPEG terminology the word “programme” is used to refer to a collection
of elementary data streams that are logically related and belong together. A
simple example is a video stream and an audio stream that together form a
conventional television programme.
Transmitter Receiver
The service content could be any mixture of video, audio and data. Delivery
could be to a conventional television set via a set-top box or even to a personal
computer.
An overview of the baseband processing and satellite channel adaptation
blocks is given in section 3. and section 4. respectively.
3. Baseband Processing
3.1. MPEG-2
The established MPEG-2 standard was adopted in DVB for the source coding
of audio and video information and for multiplexing a number of source data
streams and ancillary information into a single data stream suitable for
transmission. Consequently, many of the parameters, fields and syntax used
in a DVB baseband processor are specified in the relevant MPEG-2 standards
(See Annex C).
MPEG stands for Moving Picture Experts Group, a group of experts drawn
from industry who contribute to the development of common standards
through an ITU-T and ISO/IEC joint committee.
The MPEG-2 standards are generic and very wide in scope and consequently
some of the parameters and fields of MPEG-2 are not used in DVB. These
restrictions are described in an ETSI Technical Report (ETR 154).
Each of the video, audio and programme-related data inputs is encoded and
formatted into a Packetized Elementary Stream (PES). Thus each PES is a
digitally encoded component of a programme.
Video PES
Packetizer
Encoder
Programme Source
Transport Multiplexer
Audio PES
Packetizer To
Encoder PESs from
TS Satellite
other
PES programmes Channel
Data
Packetizer Adaptor
Encoder
Clock
Key
Timing SI/ Private
CA PES = Packetized Elementary Stream
Data PSI Data
TS = Transport Stream
SI = Service Information
PSI = Programme-Specific Information
Control/Other Data
CA = Conditional Access
3.3. Programmes
3.3.1.a Video
MPEG-2 encoding exploits the fact that picture elements that are physically
adjacent in the same frame (image) may be identical or very similar. In this
case, the picture elements are said to be “spatially correlated” and it is
assumed that the magnitude of any particular picture element can be predicted
from the nearby picture elements in the same frame. Redundant information
can therefore be removed by “intra-frame” coding. The MPEG-2 compression
algorithms employ Discrete Cosine Transform (DCT) coding techniques for
this purpose. This mathematical transformation is performed on small blocks
of 8 x 8 pixels to produce a series of coefficients. The image information tends
to be concentrated into the first few coefficients and many of the “higher
frequency” coefficients are often close to zero. Bit rate reduction is achieved
by discarding the higher frequency elements, which typically carry redundant
information that need not be transmitted. With DCT coding, the image can be
reconstructed from data corresponding to a single video frame.
Picture elements can also be correlated in time (temporal correlation), that is,
the magnitude of a particular picture element can be predicted from picture
elements in a nearby frame. “Inter-frame” coding techniques in the form of
Differential Pulse Code Modulation (DPCM) can be used to remove temporal
redundancy from the video sequence. In this case, data corresponding to more
than one video frame is required in the decoder in order to reconstruct the
image. Inter-frame coding therefore requires more decoder memory.
MPEG-2 video coding employs an adaptive combination of temporal
prediction followed by transform coding of the remaining spatial information
(hybrid DPCM/DCT).
Other techniques are employed in order to obtain better compression
performance. The resolution of the input video is often reduced by sub-
sampling prior to encoding. Clearly, this helps by reducing the number of
picture elements to be coded. Sub-sampling can also be performed in the
temporal domain in order to reduce the frame rate before coding, thus
reducing the number of picture elements that need to be processed per second.
The picture sequence is reconstructed in the decoder at the correct frame rate
by interpolation between frames.
A technique called motion compensation is also employed. This is based on
the estimation of motion between frames. If, for example, the content of a
particular video frame were simply a spatially displaced version of the content
of the previous video frame, then the image could be reconstructed from the
previous image and a “motion vector” describing the direction of movement.
A video sequence processed by an MPEG-2 encoder can consist of three types
of picture: “I”, “P” and “B”.
I (Intra) pictures are encoded without reference to other pictures in the video
sequence. They provide access points in the bit stream for random access or
switching and are thus used at the start or re-start of a decoding sequence.
However, the inherent data compression is low. “I” pictures can also be used
to limit the degradation introduced by the compression coding, at the expense
of an increase in the output bit rate.
Inter-frame Predicted (“P”) pictures are coded with reference to the nearest
previously-coded I picture or P picture, usually incorporating motion
compensation to increase coding efficiency. Since they are usually used as a
reference for prediction of future or past pictures, they cannot be used as
random access points.
Bi-directional predicted/interpolated (“B”) pictures are coded with reference
to past and future frames using motion compensation to achieve high
compression. “B” frames add complexity to the system but also produce a
significant saving in bit-rate.
The video encoding algorithm is lossy in the sense that the quality of the
decoded image can never be the same as the quality of the original image,
because some of the image information is discarded in the encoding process.
Clearly, the lower the bit rate used to transmit the video information, the more
information must be discarded and hence the lower the quality of the
reconstructed image. Thus there is a basic trade-off between multiplex
capacity (number of programmes) and video quality. Furthermore, MPEG-2
encoder manufacturers are free to determine their own mix of I, P and B
pictures to suit the application and to achieve an appropriate trade-off between
coding efficiency and complexity. The resulting image quality will therefore
depend upon the degree of sophistication of the compression technique and
the allocated transmission capacity, but also upon the complexity of the image
or video scene (i.e. the amount of temporal and spatial redundancy that can be
exploited).
SNR Allows for a layered bit stream, comprising a base layer and No
Scaleable supplementary signals. The supplementary signals improve the
SNR with respect to base layer decoding only.
A “Level” specifies the range of parameters that are supported by the encoder/
decoder (i.e. image size, frame rate and bit rates). The four MPEG-2 levels are
summarised in Table 2.
Generally speaking, MPEG-2 encoders for broadcast applications operate at
Main Profile, Main Level and are capable of encoding 4:3 format television
at a quality comparable to that achieved with existing PAL and SECAM
standards, at bit-rates from 1.5 Mbit/s to around 15 Mbit/s.
3.3.1.b Audio
The full range of bit rates for each layer is supported. The three sampling
frequencies, 32, 44.1 and 48 kHz, are supported for primary sound services.
Other sampling rates may be used for secondary sound services, but support
of these rates in the DVB IRD is optional.
3.3.1.c Data
With reference to Figure 2, it can be seen that certain non-video and non-
audio data is processed before being input to the transport multiplexer,
whereas other data, such as the Service Information, is input directly to the
transport multiplexer.
The reason for this is that some data needs to be synchronised with the video
and audio content of a specific programme. This data is encoded into
Programme Elementary Stream (PES) packets in order to take advantage of
the in-built time stamp and synchronisation facilities of the PES.
Two specific examples of this type of data are teletext and subtitling. Standard
methods exist for transporting this information in DVB bit streams and are
reviewed in the following sections.
3.3.1.c.1 Teletext
The ETSI standard EN 300 472 specifies the method by which ITU-R System
B Teletext2 may be carried in DVB bit streams.
The transport mechanism is intended to support the transcoding of the teletext
data into the Vertical Blanking Interval (VBI) of analogue video, such that it
is compatible with existing TV receivers equipped with teletext decoders. It
is specified to be capable of transmitting subtitles with accurate timing with
respect to the video (i.e. to within or near frame accuracy).
Teletext data are conveyed in Packetized Elementary Stream (PES) packets
as private data. The PES packets are in turn carried by Transport Stream
packets.
The Packet Identifier (PID) of a teletext stream associated with a particular
service (programme) is identified in the Service Information, or more
specifically in the Program Map Table (PMT) of the Program Specific
Information (PSI) for that service (see Section 3.5 for a discussion on the PSI
and the PMT).
A service may include more than one teletext data stream, provided that each
stream can be uniquely identified.
2. ITU-R Recommendation 653, also known as EBU Teletext (EBU SPB 492).
SOURCE VIDEO
fixed length
830 kbytes 830 kbytes 830 kbytes 830 kbytes Presentation Units
(uncoded pictures)
MPEG-2 Coding
variable length
Access Units
(coded pictures)
Packetization
packet header
fixed or variable
payload length PES
packets
PACKETIZED ELEMENTARY STREAM
The input to the MPEG-2 encoder is assumed to be at Main Level. From Table
2, this means that the luminance component is sampled 720 times per line.
Given that there are 576 active lines per frame, each picture (frame) consists
of 720 x 576 luminance samples. The two colour difference signals are each
sampled at half this rate, producing the same amount of data per frame for the
chrominance information. Consequently, since all samples are represented
with 8 bits, each picture in the uncompressed video input represents
approximately 830 kbytes of data.
Each picture in its uncompressed form is referred to as a Presentation Unit.
The coder compresses each presentation unit to give a coded picture. A coded
picture is known as an “Access Unit”.
The video access units are not all of the same size. The size depends upon the
degree of redundancy in the picture and upon whether they represent an “I”,
“P” or “B” picture (see Section 3.3.1.1.1). The approximate size of a video
access unit is 100 kbytes, 33 kbytes and 12 kbytes for “I”, “P” and “B”
pictures respectively, although the size can vary considerably from these
typical values due to the varying complexity of the picture.
The output of the MPEG-2 encoder is a succession of access units that
comprise a video elementary stream. Although the input to the video coder is
at a constant bit rate, the video elementary stream at the output of the coder is
not.
The next step in the creation of the MPEG-2 multiplex is to convert each
elementary stream into a Packetized Elementary Stream (PES). A Packetized
Elementary Stream consists entirely of PES packets.
The Transport Stream is intended for broadcast systems where error resilience
is one of the most important properties. It supports multiple programmes with
independent time-bases, carries an extensible description of the contents of
the programmes in the multiplex and supports re-multiplexing and scrambling
operations.
A transport stream multiplex always consists entirely of short, fixed length
transport packets. A transport packet is always 188 bytes long and comprises
a “header”, followed by “adaptation field” or a “payload”, or both.
The mapping of PES packets into transport packets is illustrated in Figure 4.
The transport packets are usually shorter in length than the PES packets of an
elementary stream. Consequently, each PES packet is divided among the
payload parts of a number of transport packets. The start of the PES packet
must occur at the start of a transport packet payload and a transport packet
may only carry data taken from one PES packet. Since a PES packet is
unlikely to exactly fill the payloads of an integer number of transport packets,
the payload of the last transport packet will be left partially empty. This
excess capacity is deliberately wasted by including an adaptation field of the
appropriate length in this transport packet.
PES Packets
fixed or variable
length
fixed length
(188 bytes)
Transport Packets
All the packetized elementary streams that are to be multiplexed together are
converted to transport packets. The resultant transport packets are output
sequentially in any order to form an MPEG-2 Transport Stream, so long as the
chronological order of packets belonging to the same elementary stream is
preserved. Transport packets carrying other information, such as Service
Information or Conditional Access data, are also added to the multiplex,
together with “null” packets that fill up any spare multiplex capacity.
The MPEG-2 PSI and the DVB-SI are based on MPEG-2 “Sections” and
“Tables”. These are briefly described in the following section.
Conditional Bouquet
Access Association
Table Table
CAT BAT
TDT RST
Time
Offset
Table
TOT
Stuffing
Table
ST
Syntax
“Short” “Long”
section_syntax_indicator 1 1
private_indicator 1 1
reserved 2 2
section_length 12 12
All MPEG-2 PSI sections and the most important DVB-SI sections follow the
long syntax.
All sections of a table carry the same table_id, an 8-bit field. In order to allow
for more than 256 tables, there is also a 16-bit table_id_extension field (long
syntax). This allows for up to 16.8 million tables, each having a length of
about 256 kbytes (or about 1 Mbyte for EITs).
The section_number and last_section_number fields of the section header
support the re-assembly of the table in the receiver. The section_number is a
simple counter to identify the place of the section within a sequence of
sections. The last_section_number identifies the last section of a table. If a
receiver has loaded all sections having the same table_id with section
numbers up to and including the last_section_number, then it knows that it
has loaded the entire table.
The version_number field is used to inform a receiver that a new version of a
database (Table) needs to be downloaded. The version_number field is
incremented each time new data is transmitted within a table. Since this field
is only five bits wide, the version number repeats every 16 tables. It is very
important that the version_number field is updated following each table
update, otherwise the receiver may become confused.
Note that a fourth MPEG-2 table is shown in Figure 5. This is the Transport
Stream Description Table (TSDT). Its purpose is to provide compatibility
with other delivery systems such as SNG (Satellite News Gathering) and
ATSC (North American television standard). It is not described further in this
document.
PID = 0
Network Information Table
(NIT)
Programme PID Value
No.
0 16
1 83
3 76
7 93
5 93
etc. etc.
Stream PID Value
(other data)
PCR 56
Programme Association Table
video 63
(PAT)
audio 1 64
audio 2 71
subtitling 95
transport packets
etc. etc.
containing the
(other data) video elementary
stream for programme
Programme Map Table no. 3
(PMT)
Unlike the MPEG-2 PSI, the DVB-SI provides information on services and
events carried by different multiplexes, and even on services and events
carried by other networks.
The DVB-SI data is structured as nine tables:
n the Bouquet Association Table (BAT),
n the Service Description Table (SDT),
n the Event Information Table (EIT),
n the Running Status Table (RST),
n the Time and Date Table (TDT),
n the Time Offset Table (TOT),
n the Stuffing Table (ST),
n the Selection Information Table (SIT), and
n the Discontinuity Information Table (DIT).
In addition, as discussed previously, the structure of the NIT is defined by
DVB.
The various DVB-SI tables are briefly described in the following sections.
Guidelines for processing the Service Information at delivery media
boundaries (e.g. from satellite to cable or SMATV systems) can be found in
ETSI Report ETR 211.
There are two basic types of event data held in the EIT: “present/following”
and “schedule”. Present/following data refers to the currently running
(“present”) event and the next (“following”) event of a given service.
Schedule data consists of a list of events that will take place at some time after
the next event. The EIT schedule tables are optional.
The present/following EIT can be used to implement a simple user interface
in the receiver. The schedule EIT can be used to implement a more complex
EPG.
The EIT can contain event information concerning the actual transport stream
(the one carrying the SDT) or other transport streams.
DVB has specified a common interface for conditional access (and other DVB
decoder applications) that is suitable for use in MultiCrypt applications. It
provides a means for the IRD to descramble programmes that have been
broadcast in parallel using different CA systems through the insertion of a
PCMCIA module into the common interface.
In SimulCrypt applications, the viewer is able to use a single CA system built
into his IRD to watch all programmes, irrespective of the CA system used to
scramble each programme. This has some security benefits with respect to
MultiCrypt because the CA module is not usually physically accessible to the
user. Furthermore, there is potentially some cost saving in the manufacture of
the IRD since the common interface need not be provided, although the cost
of implementing this interface has fallen considerably and continues to fall.
In contrast to MultiCrypt, the Simulcrypt approach requires CA system
operators to establish commercial agreements amongst themselves, which can
sometimes be difficult. It also implies dedicated IRD hardware that cannot be
upgraded or adapted for use with other CA systems and therefore may become
obsolete with time.
Both MultiCrypt and SimulCrypt have their advantages and disadvantages, as
described above. With these in mind, service providers will choose the
approach that best meets their commercial objectives. Both approaches can
and do coexist in satellite-based delivery systems.
The ETSI EN 300 421 standard specifies the modulation and channel coding
system for satellite digital multi-programme TV services, both for primary
and secondary distribution and for use in the Fixed Satellite Service (FSS) and
in the Broadcast Satellite Service (BSS) frequency bands. This is commonly
referred to as the “DVB-S” standard.
The standard is intended for Direct-To-Home (DTH) services to consumer
Integrated Receiver Decoders (IRD), as well as for reception via collective
antenna systems (Satellite Master Antenna Television (SMATV)) and at
cable television head-end stations. It can support the use of different satellite
transponder bandwidths, although a bandwidth of 33 MHz is commonly used.
All service components (“programmes”) are Time Division Multiplexed
(TDM) into a single MPEG-2 transport stream, which is then transmitted on
a single digital carrier.
3. The ETSI standard EN 300 421 gives several examples of the use of the DVB-S sys-
tem for 33 MHz and other transponder bandwidths and for different code rates.
4. This corresponds to a BER of approximately 2 x 10-4 at the output of the inner error
correction (Viterbi) decoder.
The DVB-S standard specifies the Eb/No values at which QEF quality must
be achieved when the output of the modulator is directly connected to the
input of the demodulator (i.e. in an “IF loop”). An allowance is made for
practical implementation of the modulator and demodulator functions and for
the small degradation introduced by the satellite channel. The values range
from 4.5 dB for rate 1/2 convolutional coding to 6.4 dB for rate 7/8
convolutional coding.
As mentioned previously, the inner code rate can be varied to increase or
decrease the degree of error protection for the satellite link at the expense of
capacity. Table 4 illustrates this trade off with respect to the common case of
rate 3/4 inner coding, assuming a constant transmitted symbol rate (i.e. a fixed
satellite transponder bandwidth). It shows the reduction or increase in
capacity associated with a change in the code rate and the related increase or
reduction in the Eb/No requirement. The latter is also expressed as an
equivalent increase or reduction in the diameter of the receive antenna (the
size of user’s satellite dish), all other link parameters remaining unchanged.
3/4 reference
Mux to Satellite
MPEG-2 Adaptation Channel
Outer Inter- Inner Baseband QPSK
Transport and via
Coder leaver Coder Shaping Modulator
Stream Energy Transmit
Dispersal Earth
Station
5. Contact Details
Network Development
70 rue Balard
F - 75502 PARIS CEDEX 15
France
Fax: + 33 1 53 98 40 23