International organisation for standardisation organisation internationale de normalisation
|
|  Yüklə 1.39 Mb.
|
Annex J(Informative.) Systems conformance and real-time interface J.0 Systems conformance and real-time interfaceConformance for ITU T Rec. H.222.0†|†ISO/IEC 13818-1 Program Streams and Transport Streams is specified in terms of the normative specifications in this Part of this Recommendation†|†International standard. These specifications include, among other requirements, a System Target Decoder (T-STD and P STD) which specifies the behavior of an idealized decoder when the stream is the input to such a decoder. This model, and the associated verification, do not include information concerning the real-time delivery performance of the stream, except for the accuracy of the system clock frequency which is represented by the Transport Stream and the Program Stream. All Transport Streams and Program Streams must comply with this specification. In addition, there is a real-time interface specification for input of Transport Streams and Program Streams to a decoder. This specification allows standardization of the interface between MPEG decoders and adapters to networks, channels, or storage media. The timing effects of channels, and the inability of practical adapters to eliminate completely these effects, causes deviations from the idealized byte delivery schedule to occur. While it is not necessary for all MPEG decoders to implement this interface, implementations which include the interface shall adhere to the specifications. This specification covers the real-time delivery behavior of Transport Streams and Program streams to decoders, such that the coded data buffers in decoders are guaranteed not to overflow nor underflow, and decoders are guaranteed to be able to perform clock recovery with the performance required by their applications. The MPEG real-time interface specifies the maximum allowable amount of deviation from the idealized byte delivery schedule which is indicated by the Program Clock Reference (PCR) and System Clock Reference (SCR) fields encoded in the stream. Annex K(Informative.) Interfacing Jitter-Inducing Networks to MPEG-2 Decoders K.0 Introduction In this Annex the expression system stream will be used to refer to both ITU T Rec. H.222.0†|†ISO/IEC 13818 1 Transport Streams and ITU T Rec. H.222.0†|†ISO/IEC 13818 1 Program Streams. When the term STD is used, it is understood to mean the P STD (Program System Target Decoder) for Program Streams and the T-STD (Transport System Target Decoder) for Transport Streams. The intended byte delivery schedule of a system stream can be deduced by analyzing the stream. A system stream is compliant if it can be decoded by the STD, which is a mathematical model of an idealized decoder. If a compliant system stream is transmitted over a jitter-inducing network, the true byte delivery schedule may differ significantly from the intended byte delivery schedule. In such cases it may not be possible to decode the system stream on such an idealized decoder, because jitter may cause buffer overflows or underflows and may make it difficult to recover the time base. An important example of such a jitter-inducing network is ATM. The purpose of this Annex is to provide guidance and insight to entities concerned with sending system streams over jitter-inducing networks. Network specific compliance models for transporting system streams are likely to be developed for several types of networks, including ATM. The STD plus a real-time interface definition can play an integral role in defining such models. A framework for developing network compliance models is presented in K.2. Three examples of network encoding to enable the building of jitter-smoothing network adapters are discussed in K.3. In the first example a constant bitrate system stream is assumed and a FIFO is used for jitter smoothing. In the second example the network adaptation layer includes timestamps to facilitate jitter smoothing. In the final example a common network clock is assumed to be available end-to-end, and is exploited to achieve jitter smoothing. K.4 presents two examples of decoder implementations in which network-induced jitter can be accommodated. In the first example, a jitter-smoothing network adapter is inserted between a network's output and an MPEG-2 decoder. The MPEG-2 decoder is assumed to conform to a real-time MPEG-2 interface specification. This interface requires an MPEG-2 decoder with more jitter tolerance than the idealized decoder of the STD. The network adapter processes the incoming jittered bitstream and outputs a system stream whose true byte delivery schedule conforms to the real-time specification. Example one is discussed in K.4.1. For some applications the network adapter approach will be too costly because it requires two stages of processing. Therefore, in the second example the dejittering and MPEG-2 decoding functions are integrated. The intermediate processing of the jitter-removal device is bypassed, so only a single stage of clock recovery is required. Decoders that perform integrated dejittering and decoding are referred to in this Annex as integrated network-specific decoders, or simply integrated decoders. Integrated decoders are discussed in K.4.2. In order to build either network adapters or integrated decoders a maximum value for the peak-to-peak network jitter must be assumed. In order to promote interoperability, a peak-to-peak jitter bound must be specified for each relevant network type. K.1 Network compliance models One way to model the transmission of a system stream across a jitter-inducing network is shown in Figure K 1. 
Figure K-1 -- Sending system streams over a jitter-inducing network The system stream is input to a network-specific encoding device that converts the system stream into a network specific format. Information to assist in jitter removal at the network output may be part of this format. The network decoder comprises a network-specific decoder and an ITU T Rec. H.222.0†|†ISO/IEC 13818 1 decoder. The ITU T Rec. H.222.0†|†ISO/IEC 13818 1 decoder is assumed to conform to a real-time interface specification, and could have the same architecture as the STD with appropriate buffers made larger to provide more jitter tolerance. The network-specific decoder removes the non- ITU T Rec. H.222.0†|†ISO/IEC 13818 1 data added by the network-specific encoder and dejitters the network's output. The output of the network-specific decoder is a system stream that conforms to the real-time specification. A network target decoder (NTD) can be defined based on the above architecture. A compliant network bitstream would be one that was able to be decoded by the NTD. A network decoder would be compliant provided it could decode any network bitstream able to be decoded by the NTD. A real network decoder might or might not have the architecture of the NTD. K.2 Network specification for jitter smoothing In the case of constant bit rate system streams, jitter smoothing can be accomplished with a FIFO; additional data that provides specific support for dejittering is not required in the network adaptation layer. After the bytes added by the network encoding are removed, the system stream data is placed in a FIFO. A PLL keeps the buffer approximately half full by adjusting the output rate in response to changes in buffer fullness. In this example the amount of jitter-smoothing achieved will depend on the size of the FIFO and the characteristics of the PLL. Figure K 2 illustrates a second way to accomplish jitter smoothing; in this example timestamp support from a network adaptation layer is assumed. Using this technique both constant bit rate and variable bit rate system streams can be dejittered. 
Figure K-2 -- Jitter smoothing using network-layer timestamps Assume the network adapter is designed to compensate for a peak-to-peak jitter of J seconds. The intended byte delivery schedule is reconstructed using clock reference samples (CRs) taken from a time clock (TC). The CRs and the TC are analogous to PCRs and the STC. The network data packet (NDP) encode converts each system stream packet into a network data packet (NDP). The network data packets contain a field for carrying CR values, and the current value of the TC is inserted into this field as the NDP leaves the NDP encoder. The network transport packetization (NXP) function encapsulates the NDPs into network transport packets. After transmission across the network, the CRs are extracted by the NDP decoder as the NDPs enter the NDP decoder. The CRs are used to reconstruct the TC, for example by using a PLL. The first MPEG-2 packet is removed from the dejittering buffer when the delayed TC (TCd) is equal to the first MPEG-2 packet's CR. Subsequent MPEG-2 packets are removed when their CR values equal the value of the TCd. Ignoring implementation details such as the speed of the TC clock recovery loop and the spectral purity of the TC, the size of the dejittering buffer depends only on the maximum peak-to-peak jitter to be smoothed and the largest transport rate that occurs in the system stream. The dejittering buffer size,Bdj, is given by 
where Rmax is the maximum data rate of the system stream in bits per second. When packets traversing the network experience the nominal delay, the buffer is half full. When they experience a delay of J/2 seconds, the buffer is empty, and when they experience a delay (advance) of -J/2 seconds the buffer is full. As a final example, in some cases a common network clock will be available end-to-end, and it may be feasible to lock the system clock frequency to the common clock. The network adapter can smooth jitter with a FIFO. The adapter uses PCRs or SCRs to reconstruct the original byte delivery schedule. |