The Research On The Quality Control For Network Live Video Broadcast

With the rapid development of Internet and multimedia technologies, live video broadcast technology has been applied widely. In order to keep playback quality of client side acceptable, network live video broadcast system should be adaptive, real-time and steady. But, network video broadcast today is plagued by variability in throughput, packet loss and delay due to network congestion and the heterogeneous infrastructure, which impact the playback quality badly. In addition, network bandwidth becomes more and more limited upon the rapid increasing of Internet user. It is much more important to control the quality of live video streaming.In this paper, several algorithms have been proposed to keep the playback quality acceptable through buffering and flow control technologies that enable streaming live video over networks reliably. The main work we have done can be summarized as follows.1 We present an end-to-end transport architecture based on RTP protocol that can make live video streaming more effective.2 We study the packet losses problem caused by instantaneous sending rate and network congestion deeply, and give out a server buffering method which can make a better solution for this problem.3 In order to improve those existing client buffering methods, we propose an Adaptive Management Strategy (AMS), which provides a delay-sensitive solution for client buffering. Experiments demonstrate that AMS can achieve a better trade-off between packet losses and buffering delay.4 Due to the imprecise prediction of network traffic, existing flow control methods can not adapt to the dynamics of network. Aiming at solving these problems, we propose an Adaptive Rate Transmission Strategy (ARTS), which is TCP-friendly and can provide a better solution for real-time video transmission.5 Based on our research, we develop an actual network video broadcast system through DirectShow technology. The realization of quality control for this system is discussed in details. Experimental results demonstrate that the system we developed can obtain a stable transmission rate. The playback of live video streaming can be synchronized for all clients. And the broadcast delay can be limited in 1s.