Chapter 7. MPEG-2 SYSTEM LAYER
7.0 Introduction
The compressed image and video data is stored and transmitted
in a standard format known as a compression stream. The discussion in this
section will be restricted exclusively to the presentation of the video compression
stream standards associated with the MPEG-2 systems layer: elementary stream
(ES), packetized elementary stream (PES), program stream (PS), and transport
stream (TS)
The MPEG-2 systems layer is responsible for the integration and synchronization
of the elementary streams (ES): audio and video streams, as well as an unlimited
number of data and control streams that can be used for various applications
such as subtitles in multiple languages. This is accomplished by first packetizing
the ESs thus forming the packetized elementary streams (PES). These PESs contain
timestamps from a system clock for synchronization.
Figure 1:
MPEG-2 Audio and Video Systems Layer
The PESs are subsequently multiplexed to form a single output stream
for transmission in one of two modes: program stream (PS) and transport stream
(TS). The PS is provided for error-free environments such as storage in CD-ROM.
It is used for multiplexing PESs that share a common time-base, using long
variable-length packets.[1] The TS is designed
for noisy environments such as communication over ATM networks. This mode
permits multiplexing streams (PESs and PSs) that do not necessarily share
a common time-base, using fixed-length (188 bytes) packets. An example of
the MPEG-2 systems layer illustrating the multiplexing of the packetized audio
and video elementary streams is depicted in Fig. 2
7.1 MPEG-2 Elementary
Stream
As indicated earlier, MPEG-2 systems layer supports an
unlimited number of elementary streams (ES). Our focus is centered on the
presentation of the ES format associated with the video stream. The structure
of the video ES format is dictated by the nested MPEG-2 compression standard:
video sequence, group of pictures (GOP), pictures, slices, and macroblocks.
The video ES is defined as a collection of access units (pictures) from one
source. An illustration of the video ES format is depicted in Fig. 2. A corresponding
glossary of the video ES format is provided in Table 1. Note that the unshaded
segment of the video ES format presented in Fig. 2 is used to denote that
any permutation of the fields within this segment can be repeated as specified
by the video compression standard.
|
SH |
SE |
EUD0 |
GOPH |
EUD1 |
PH |
PCE |
EUD2 |
PDS |
SDMB |
SE |
Figure 2: Video Elementary Stream Format
|
Abbr |
Function |
|
SH |
Sequence Header |
|
SE |
Sequence Extension |
|
EUD0 |
Extension and User Data 0 |
|
GOPH |
Group Of Picture Header |
|
EUD1 |
Extension and User
Data 1 |
|
PH |
Picture Header |
|
PCE |
Picture Coding Extension |
|
EUD2 |
Extension and User
Data 2 |
|
PDS |
Picture Data containing Slices |
|
SDMB |
Slices Data containing Macro-Blocks |
|
SE |
Sequence End |
Table 1: Video Elementary Stream Format Glossary
7.2 MPEG-2 Packetized
Elementary Stream
The MPEG-2 systems layer packetizes all ESs---audio, video, data, and control streams---thus forming the packetized elementary streams (PES). Each PES is a variable-length packet with a variable format that corresponds to a single ES.
The elementary stream clock reference (ESCR) indicates the intended
time of arrival of the packet at the system target decoder (STD). The rate
at which the STD receives the PES is indicated by the elementary stream rate
(ESR). Error checking is provided by the PES cyclic redundancy check (PESCRC).
The pack header field (PHF) is
a PS pack header. The program packet sequence counter (PPSC) indicates the
number of system streams. The STD buffer size is specified by the P-STD buffer
(PSTDB) field.
A nested representation of the
PES header is depicted in Fig. 3. The corresponding glossary of the PES header
is provided in Table 2. Note that the unshaded boxes presented in Fig. 3are
used to represent optional fields in the PES header.
|
PSCP |
SID |
PESPL |
OPESH |
SB |
Primary PES Header
|
10 |
SC |
P |
DAI |
C |
O/C |
Flags (7) |
PESHDL |
OF1 |
Optional PES Header
|
PTS/ DTS |
ESCR |
ESR |
DSMTM |
ACI |
PESCRC |
PESE |
Optional Fields 1
|
Flags (5) |
OF2 |
PES Extension
|
PESPD |
PHF |
PPSC |
PSTDB |
PESEF |
Optional Fields 2
Figure 3: PES Header
|
Abbr |
Function |
|
PSCP |
Packet Start Code Prefix |
|
SID |
Stream Identification |
|
PESPL |
Packetized Elementary Stream Packet Length |
|
OPESH |
Optional Packetized Elementary Stream Header |
|
SB |
Stuffing Bytes |
|
SC |
Scrambling Control |
|
P |
Priority |
|
DAI |
Data Alignment Indicator |
|
C |
Copyright |
|
O/C |
Original/Copy |
|
Flags (7) |
7 Flags |
|
PESHDL |
Packetized Elementary
Stream Header Data Length |
|
OF1 |
Optional Fields 1 |
|
PTS/DTS |
Presentation Time-Stamps
/ Decoding Time-Stamps |
|
ESCR |
Elementary Stream Clock
Reference |
|
ESR |
Elementary Stream Rate |
|
DSMTM |
DSM Trick Mode |
|
ACI |
Additional Copy Information |
|
PESCRC |
Packetized Elementary Stream Cyclic Redundancy Check |
|
PESE |
Packetized Elementary Stream Extension |
|
Flags (5) |
5 Flags |
|
OF2 |
Optional Fields 2 |
|
PESPD |
Packetized Elementary
Stream Private Data |
|
PHF |
Pack Header Field |
|
PPSC |
Program Packet Sequence Counter |
|
PSTDB |
P-STD Buffer |
|
PESEF |
Packetized Elementary Stream Extension Field |
Table 2: PES Header Glossary
7.3 MPEG-2 Program Stream
A program stream (PS) multiplexes several PESs, which
share a common time-base, to form a single stream for transmission in error-free
environments. The PS is intended for the storage and retrieval of programs
from digital storage media such as CD-ROM. The PS uses relatively long variable-length
packet. For a more detailed presentation of the MPEG-2 PS refer to X.
7.4 MPEG-2 Transport
Stream
A transport stream (TS) permits multiplexing streams (PESs
and PSs) that do not necessarily share a common time-base for transmission
in noisy environments. The TS is designed for broadcasting over communication
networks such as ATM networks. The TS uses small fixed-length packets (188
bytes) that make them more resilient to packet loss or damage during transmission.
The TS provides the input to the transport layer in the OSI reference model.[3]
The TS packet is composed of a four-byte header followed by 184 bytes
shared between the variable-length adaptation field and the TS packet payload.
An illustration of the TS header is depicted in Fig. 4. The corresponding
glossary of the TS header is provided in Table 3. Note that the unshaded box
appearing in Fig. 4 is used to represent the optional adaptation field (AF).
|
SB |
TEI |
PUSI |
TSC |
TP |
PID |
AFC |
CC |
AF |
Figure 4: TS Header
|
Abbr |
Function |
|
SB |
Synchronization Byte |
|
TEI |
Transport Error Indicator |
|
PUSI |
Payload Unit Start Indicator |
|
TSC |
Transport Scrambling Control |
|
TP |
Transport Priority |
|
PID |
Packet Identifier |
|
AFC |
Adaptation Field Control |
|
CC |
Continuity Counter |
|
AF |
Adaptation Field (Optional) |
Table 3: TS Header Glossary
The TS header includes a synchronization
byte (SB) designed for detection of the beginning of each TS packet. The transport
error indicator (TEI) points to the detection of an uncorrectable bit error
in this TS packet. The payload unit start indicator (PUSI) is used to ascertain
if the TS payload contains PES packets or program-specific information (PSI).
The packet ID (PID) identifies the type and source of payload in the TS packet.
The presence or absence of the adaptation field (AF) and payload is indicated
by the adaptation field control (AFC). The continuity counter (CC) provides
the number of TS packets with the same PID, which is used to determine packet
loss.
The optional adaptation field (AF) contains additional information
that need not be included in every TS packet. One of the most important fields
in the AF is the program clock reference (PCR). The PCR is a 42-bit field
composed of a 9-bit segment incremented at 27 MHz as well as a 33-bit segment
incremented at 90 KHz.[4]
The PCR is used along with a voltage controlled oscillator as a time reference
for synchronization of the encoder and decoder clock.
A PES header must always follow
the TS header and possible AF. The TS payload may consist of the PES packets
or PSI. The PSI provides control and management information used to associate
particular ESs with distinct programs. A program is once again defined as
a collection of ESs that share a common time-base. This is accomplished by
means of a program description provided by a set of PSI associated signaling
tables (AST): program association tables (PAT), program map tables (PMT),
network information tables (NIT), and conditional access tables (CAT). The
PSI tables are sent periodically and carried in sections along with cyclic
redundancy check (CRC) protection in the TS payload.
|
I Picture |
P Picture |
P Picture |
Elementary Stream (ES)
|
PES Header |
I Picture |
|
PES Header |
P Picture |
Packetized Elementary Stream (PES)
0 187B 0
187B
0
187B
|
TS Header |
PES Data |
|
TS Header |
PES Data |
|
TS Header |
PES Data |
Stuffing |
Transport
Stream (TS)
Figure 5: TS Packets
An example illustrating the formation of the TS packets is depicted in Fig. 5. The choice of the size of the fixed-length TS packets---188 bytes---is motivated by the fact that the payload of the ATM Adaptation Layer-1 (AAL-1) cell is 47 bytes. Therefore, four AAL-1 cells can accommodate a single TS packet. A detailed discussion of the mapping of the TS packets to ATM networks is presented in the next section.