Revert to Section 3.9
4.1 Available Structures
ATM technology was developed to solve the multiplex and transport problems found in the communications world of telephony. As specified in ITU/TS this is a structure in which any type of information is encapsulated in cells of 53 bytes. The first 5 bytes (the header) contain the multiplexing information and the last 48 bytes (the payload) contain the user information. In order to ensure end-to-end information and time transparencies, an ATM adaptation layer (AAL) has been defined on top of the ATM layer. Different types of AAL can be specified to cover the full range of services to be supported.
This approach could also be adapted for digital terrestrial television broadcasting (DTTB). A similar packet structure can be adopted, the functionalities and structure of the cell header being optimized for the DTTB environment. The addition of some kind of recurrent frame structure to improve the multiplex performance in poor error conditions may also be needed.
The Moving Picture Experts Group (MPEG) of ISO/IEC has produced a multiplex structure that could also be used for DTTB. In North America, the transport mechanism for the Advanced Television System Committee Standard is a subset of the MPEG-2 System Transport Stream Syntax.
In Europe, digital multi-programme TV systems have been developed by the DVB project, for Satellite, CATV and SMATV (Satellite Master Antenna Television systems) applications. These systems make use of MPEG-2 video and audio coding methods, as well as of Transport Stream Multiplexing. To achieve the maximum commonality over the various media, MPEG source coding and transport multiplexing methods will be adopted by the DVB project also for the DTTB system under development.
The MPEG-2 system packet multiplex structures were specifically tailored to the needs of broadcast video, audio and data signals with consideration of compatibility with ATM structures included. The MPEG-2 systems packet structure consists of 188 bytes comprised of 4 header bytes and 184 payload bytes. This packet size was designed to be encapsulated within four ATM cells as four 47-byte payloads (4 x 47 = 188) leaving space for 1 ATM AAL byte per ATM cell. The MPEG-2 system can transport data with a lower overhead than the ATM system. Overhead is an important consideration in the highly constrained DTTB environment.
The Japanese Administration proposed application of ISDB to digital broadcasting, noting the following inherent characteristics:
Satellite, terrestrial and cable digital broadcasting systems are being developed aiming at the maximally common specification based on the concept of ISDB.
4.2 Multiplexing of video, audio, and data
The various component video, audio, and data element bit-streams need to be multiplexed together to form the complex signal ensemble in a DTTB system. In addition, various error protection strategies can be employed at this stage of processing to enhance the ruggedness of the multiplexed data. Randomization of the digital data and data interleaving are possible techniques that can be used so that bursts of channel bit errors can be treated as uncorrelated bit errors by the forward error correction codecs. In addition, synchronization bit sequences can be inserted at intervals to mark these boundaries and provide appropriate video, audio, data/text and control data streams to the processor.
One approach is to arrange transmission bytes in a structure analogous to the line and field structure of existing analogue TV signals. This may also result in a signal structure containing recurring sequences that can be used for synchronization and also as "training signals" for ghost-canceler or channel-equalizer systems.
An alternative approach could be the use of a cell-relay-based data-transport layer that supports the prioritized delivery of video data, thus providing graceful degradation of services under impaired channel conditions. The cell relay might also provide logical synchronization that is essential for reliable delivery of variable-length-coded compressed video in the presence of transmission errors. This data transport protocol also offers service flexibility for a wide mixture of video, audio, and auxiliary data services. The transport processor asynchronously multiplexes the payload data with different priorities into basic transport units called cells. A cell resembles a data packet in conventional packet networks in modern data communication. It has a header and a trailer enclosing a payload area. Each cell has a fixed size and its own error control bits. It is noteworthy that the cell format can be transcodable to B-ISDN (broadband integrated services digital network), thus providing a path for development of future information services.
Use of "headers and descriptors" within the data stream is seen by some as a useful approach that facilitates data processing.
4.2.2 Program versus Transport Stream multiplexing
System Level Multiplexing Approaches
In general there are two approaches to multiplexing elementary bit streams from multiple applications on to a single channel. One approach is based on the use of fixed length packets and the other on variable length packetisation. As illustrated in Fig. 25, the video and audio elementary bit streams in both cases are generally first formed into variable length PES (packet elementary streams) packets (although it should be noted that some applications produce fixed length PES packets). The process of generating the multiplexed bit streams for the two approaches involves a difference in processing only at the final multiplexing stage.
Figure 26 provides examples of bit streams for both the program and transport stream approaches to demonstrate their differences. As shown in Fig. 26, in the program stream approach, PES packets from various elementary bit streams are multiplexed by transmitting the bits for the complete PES packets in sequence, resulting in a sequence of variable length packets on the channel.
In contrast, the transport stream approach, as shown in Fig. 26, the PES packets (including the PES headers) are transmitted as a payload of fixed length transport packets. Each transport packet is preceded by a transport header which includes information for bit stream identification. Each PES packet for a particular elementary bit stream occupies a variable number of transport packets, and data from various elementary bit streams are generally interleaved with each other at the transport packet layer. Identification of each elementary bit stream is facilitated by the data in the transport headers. New PES packets always start a new transport packet and stuffing bytes are used to fill packets with partial PES data.
The two multiplexing schemes are motivated by different application requirements. Transport streams are appropriate for environments where errors and data loss events are likely, including certain storage media and transmission on noisy channels. Program streams are appropriate for relatively error-free media such as CD-ROMs. Errors or loss of data within PES packets can potentially result in complete loss of synchronization in the decoding process. The program stream approach is used when requirement for compatibility with MPEG-1 is stipulated.
It should be noted, however, that in general, the program and transport data streams both address the same general layers of protocol functionality, and therefore it does not make sense to carry a program bit stream within a transport bit stream or vice-versa. Transcoding between the two formats is feasible and one could build an interface between them.
Continue to Section 4.2.3
Return to DTTB Tutorial Table Of Contents
Return to Tutorial Index Page