Real-time Streaming Media Content Distribution System Optimization

With the wide spread of computer and network, the development of audio and video codec technology and the increment of life requirement, streaming media technology has been a focus in both academic circles and industrial world. As streaming media requires high bandwidth, long duration and strict QoS, it poses a great challenge for researchers to develop a large-scale capable, cost-effective, QoS guaranteeing, ISPs-friendly Streaming Content Delivery System.This dissertation is a further research on the subject of live streaming content delivery system; optimizing the distribution of content delivery performance and wireless transmission. The contributions of the dissertation are:1. For the performance challenge, which is generated by time-shifting operations based on the original live streaming content delivery system, a novel control theory based schedule algorithm called CLEAR SCHEDULE is proposed. The algorithm can effectively restrain the disturbance, help live streaming content delivery system to ensure the video quality under time-shifting conditions. Tests show that the live streaming content delivery system using CLEAR SCHEDULE algorithm has achieved good results in aspects of response speed, buffer control and rate smoothing.2. A media encapsulation format, which can optimize the fast forwarding & fast back-warding operations, is brought forward after the analysis of the bottlenecks of system I/O. This media encapsulation format support existing video encoding formats (such as MPEG4, H.264), can speed up the key frame search and read speed, reduce the unwanted data, and improve disk utilization.3. After analysis, UDP protocol is abandoned in favor of the RTP-on-TCP transport mode based on the characteristics of wireless multimedia transmission. A Multi-TCP transport mode has been designed and implemented. Experiments show it is better than the original UDP transfer mode in the wireless environment.