MPEG-4 FGS Receiver-Driven Layered Reliable Multicast

• Main Authors: Jen-Yu Yu, 游人諭
• Other Authors: Suh-Yin Lee
• Format: Others
• Language: en_US
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/44707575913215103910

碩士 === 國立交通大學 === 資訊工程系 === 90 === Traditionally, IP multicast [1] video streaming is an efficient transmission method of sending data to a group, but the packets are sent unreliably and the single target rate cannot meet all receivers’ requirements. Because the source-based rate-adaptation methods perform poorly in multicast transmission, one approach for multicast rate adaptation is to combine the layered source coding algorithms, such as SNR scalability [3] and temporal scalability [43] [44], with a multicasting transmission system, namely, Receiver-driven Layered Multicast (RLM) [2]. However, the actual available bandwidth varies over time, so the packet losses cannot be avoided and may impact the video quality seriously. Comparing with single layered coding [48] and scalable layered video coding, the advantage of MPEG-4 FGS (Fine Granularity Scalability) [9] coding scheme for Internet video streaming is more resilient to IP packet-losses and provides partial quality enhancement proportional to the number of bits which is decoded for each frame. Besides, some reliable multicast schemes, such as RMTP (Reliable Multicast Transport Protocol) [28] [29] and SRM (Scalable Reliable Multicast) [19] [30], were proposed to perform the loss recovery and feedback for reliability. In this thesis, we propose a new integrated reliable multicast streaming system, namely, MPEG-4 FGS Receiver-driven Layered Reliable Multicast (RLRM) adopting our proposed reliable multicast algorithms and exploring the advantages of RMTP, RLM, and MPEG-4 FGS. The characteristics of RLRM include: (1) a new application-level protocol to support multi-channel video transmission and loss recovery, (2) a buffer management for multi-channel multicast, and (3) a new scalable loss recovery mechanism for achieving quickly loss recovery and efficient bandwidth utilization.