Maintaining video quality and optimizing video delivery over the bandwidth constrained DSL last mile through intelligent packet drop

• Published in: Bell Labs technical journal 2008-03, Vol.13 (1), p.53-68
• Main Authors: Van Caenegem, Tom N. M., Struyve, Kris O., Laevens, Koen, De Vleeschauwer, Danny, Sharpe, Randy
• Format: Article
• Language: English
• ISSN: 1089-7089; 1538-7305
• DOI: 10.1002/bltj.20282

The technological challenges for an Internet Protocol television (IPTV) service provider to bring high quality video services into the home through a dedicated bandwidth limited digital subscriber line (DSL) access link are significant. However, at the same time, the dedicated DSL link provides certain opportunities as it allows for service customization on the network level, without impacting any other customers. In this paper, the concept of intelligent packet drop inside a digital subscriber line access multiplexer (DSLAM) is introduced. Intelligent packet drop makes use of the fact that some packets containing video information (e.g., I-picture or P-picture) are more important than others (e.g., B-picture), and this importance level can be indicated in the packet header. When congestion in the DSLAM occurs, the low priority packets are preferentially dropped. Provided some (simple) picture loss concealment is performed in the set-top box, intelligent packet drop will not significantly impact the quality of experience (QoE), as evidenced by subjective tests that have been conducted. Simulations have shown that the intelligent packet drop in a DSLAM enables the IPTV service provider to increase the number of customers eligible for a given IPTV service package delivered over the DSL link. It also enables fast channel change (FCC) service without dedicated bandwidth overhead.