Priority-Aware FEC Coding for High-Definition Mobile Video Delivery Using TCP

Abstract
High-Definition (HD) wireless video has already dominated popular multimedia applications to provide content-rich mobile streaming services. Transmission Control Protocol (TCP) is pervasively employed as the transport-layer solution in video communication systems to achieve firewall traversal and network-friendliness. However, it is severely challenging to effectively deliver HD mobile video using TCP over wireless networks: 1) HD live video streaming is featured by high transmission rate and stringent delay constraint; 2) wireless networks are bandwidth-limited and error-prone; and 3) the congestion control and data retransmission mechanisms in TCP may result in frequent throughput fluctuations and deadline violations. To address these critical issues, this research presents a Forward Error Correction (FEC) coding scheme dubbed PATON (Priority-Aware and TCP-Oriented codiNg). First, we develop an analytical model of FEC coding-based real-time video communication with TCP over wireless networks. Second, we propose a heuristic solution for prioritized frame selection, FEC redundancy adaptation, and packet size adjustment to maximize real-time video quality. The proposed PATON is able to effectively leverage the frame priority and TCP connection state to minimize the video distortion. We conduct the performance evaluation through extensive semi-physical emulations in Exata involving HD video streaming encoded with H.264 codec. Compared with the existing FEC coding schemes, PATON achieves appreciable improvements in terms of video Peak Signal-to-Noise Ratio (PSNR), end-to-end delay, ratio of lost frames, and goodput.
Funding Information
  • National Natural Science Foundation of China (NSFC) (61132001)
  • National High-tech R& D Program of China (863 Program) (2013AA102301)
  • New Century Excellent Talents in University (NCET-11-0592)
  • National Grand Fundamental Research 973 Program of China (2013CB329102)

This publication has 43 references indexed in Scilit: