Sunday, January 18, 2009

SVC, MANE, and the Future of the Internet

What have Scalable Video Coding (SVC), Media-Aware Network Elements (MANEs) and the Future of the Internet in common? Well, in this article, I'm trying to highlight this...

Scalable Video Coding

Scalability in video coding has a long history which started with MPEG-2 (Wikipedia) introducing temporal, spatial, signal-to-noise ratio (SNR) scalability, and data partitioning. However, as it successor MPEG-4, it lacks of coding efficiency compared to a single layer coded bitstream. With the emergence of the Advanced Video Coding (AVC) standard - a joint development between ISO/MPEG and ITU-T (H.264) referred to as Joint Video Team (JVT) - that gained momentum (i.e., AVC is THE state-of-the-art video coding standard with which one has to compare), the JVT tried it again and started working on Scalable Video Coding with the aim to introduce at most 10% coding overhead ... and this goal has been reached! SVC offers three (main) scalability dimensions:
  • temporal: frames per second, e.g., 30fps -> 15 fps
  • spatial: resolution, e.g., HTDV (1920x1080) -> SDTV (720×576)
  • SNR: quality, e.g., 32dB -> 28 dB
The main advantage of SVC is that it allows for extraction of low-quality (i.e., temporal, spatial, SNR) bitstreams by simply discarding portions of the original bitstream whereby the base layer is always compliant to AVC. In the past, we've done some measurements concerning the adaptation of SVC bitstreams utilizing a generic vs. specific approach.

Media-Aware Network Element

A Media-Aware Network Element (MANE) as defined by the IETF RFC3984 is referred to as a network element, such as a middlebox or application layer gateway that is capable of parsing certain aspects of the RTP payload headers or the RTP payload and reacting to the contents. MANEs - in its current state - are tightly aligned with SVC and, thus, allow the adaptation of SVC bitstreams within the network as shown in the figure below (taken from here).
The current status of the RTP payload format for SVC bitstreams can be found here. With SVC and its RTP payload format, it's very simple to deploy an adaptation engine within the network (i.e., the MANE) that adapts the video content according to different terminal capabilities and network characteristics (e.g., HDTV, SDTV, mobile devices).

The Future Internet

The term "Future Internet" comes from the future Internet conference that was held in spring 2008 in Bled, Slovenia and refers to the European Future Internet Portal which serves as the central forum for European activities and discussions on the future of the Internet. Within the Bled Declaration the main challenges towards the Future Internet have been highlighted, among them capabilities for supporting the creation, sharing, locating and delivery of new-media content which indeed is the result of research and development in the area of SVC and MANE respectively. EC-funded projects that are active supporters of the Bled Declaration and utilize SVC and MANE respectively are, for example, P2P-Next and SEA. However, there's still a long, difficult path ahead with a lot of challenges to be addressed, specifically in the area of Quality of Servce/Experience and, thus, adaptation of multimedia content (but that's another story...).

The next event is the Future of the Internet conference in Prague, in May 2009.

No comments: