MPEG要做新的媒体传输标准了

 

MPEG终于觉得他的13818-1过时了，在这个ALL OVER IP的时代，使用了20多年的TS和PS格式没落了，而RTP则风光无限。MPEG在09年2月的Lausanne会议上提出要做一个新的媒体传输标准"Modern MPEG Transport (MMT)"，哦，还只是个不太明确的需求。期待4月份的会议能有一些比较明确的目标。

摘录一下MPEG的原文吧：

Five areas have been identified so far:

• Transport- and file format friendly stream format: Streams are often sourced from files, and recorded into them.  It should be possible to build software that is aware of the structure of both a stream and file format, and that is able to do lossless conversion between the two.  Such software ideally is only aware of the stream and file level structures and does not need to be updated for new media types, annotation types, coding (codec) types, and so on.  For the MPEG community, lossless conversion between a stream format and the 'MP4' file format is clearly desirable.  The MP4 format, by itself, is not a stream format;  for example, its main feature supporting incremental files explicitly forbids introducing new sub-streams or new coding formats for existing sub-streams.
• Cross layer optimization between video and transport layer: As multimedia contents became one of the most important data delivered over Internet, there have been many efforts to define a transport protocols supporting efficient delivery of multimedia bitstream. A number of RTP payload formats have been defined to include some mechanisms for fragmentation and/or augmentation to adopt the variation of underlying delivery layers and to maximize the utilization of the network resources. However, those standards have lots of limitations because the compression standard and the delivery standard are developed separately in a disjoint manner. So, it is foreseen that the joint development of the structure of video bitstream and development of the delivery protocol to carry that bitstream would improve this situation.
• Error resilience for MPEG streams: MPEG standards define the decoding of conforming bitstreams. During the development phase, typical errors introduced by the communication channel and there impact on video quality are not studied thoroughly. Given the complexity of recently developed bitstream structures for SVC and MVC, efficient methods for error resilience and error concealment should be considered when developing future standards. Obviously, error concealment and resilience is not necessarily an area for standardization, but the structure of the bitstream and its packetization should provide efficient hooks to implement the desired features.
• Conversion between transport mechanisms: In 13818-1 MPEG defined 2 transport mechanisms: Transport Stream and Program Stream. The conversion between the two is not always straightforward. MMT should allow for easy conversion between legacy transports and MMT
• Content adaptation to different networks: In the today’s media landscape the term ‘convergence’ is becoming ubiquitous and is often referred in the context of multi-play services which enables the provisioning of multiple services by organizations that traditionally only offered one or two of them (e.g., triple-play: TV, Internet and telephony). However, one issue within such scenarios is still omnipresent which is the so-called ‘last mile’ problem that is referred to as the part of the network used by the customer to access the service. The diversity of these access networks is ranging from wired (e.g., xSDL, cable, FTTH) to wireless (e.g., WiFi, WiMAX, UMTS, LTE Advanced) infrastructures with heterogeneous characteristics, e.g., in terms of Quality of Service (QoS) support. Additionally, the core networks are mainly IPv4-based which offer only limited QoS capabilities, e.g., through the deployment of DiffServ or an underlying MPLS infrastructure. Although IPv6 offers intrinsic QoS support it lacks of widespread adoption on the market. This calls for content-aware networks enabling QoS support at the transport layer and above which allows for multimedia content adaptation to the characteristics of the underlying networking technologies and infrastructures.