Study On Key Techniques Of Real-Time Scalable Video Streaming System

Recently, streaming media has been widely applied in various areas, such as video conference, distance learning, real-time internet video streaming and so on. With the improvement of video compression technology, computer and network technology, streaming media technology has been developed rapidly. Development of streaming media technology promotes the applications of streaming media widely. Therefore, it is of great significance to study streaming media technology.However, the current internet is a best-effort work where loss, variable delays and bandwidth fluctuation occur during data delivery in a streaming session. So, the traditional non-scalable streaming cannot adapt to variation of available network bandwidth. Moreover, the diversities of terminals also challenge the popularization and development of streaming media. So, a new coding method should be adopted to provide network adaptive streaming service. The scalable video coding (SVC) can represent video bitstreams in different scalable layers of temporal, spatial and quality for flexible adaption to terminal and network characteristics. An adaptive network delivery strategy based scalable video coding is proposed in this paper, and validated in a real-time video streaming system based on Darwin Streaming ServerThis thesis focuses on the layers extraction strategy of scalable video stream according to different clients and network bandwidth under the heterogeneous network.Firstly, It is greatly significant for improving the QoS that how to know the network condition exactly and in time in a streaming media system. Congestion control is the technology that estimates the available network bandwidth. However, most of the current congestion control mechanisms are based on the algorithm of additive increase multiplicative decrease (AIMD). Darwin streaming server is the open source version of Apple's QuickTime streaming server, and supports various protocols and platforms. This thesis analyzes the strategy of congestion control of Darwin Streaming Server, using TCP Westwood (TCPW) to estimate the bandwidth, an improved congestion control strategy is proposed, using the estimate bandwidth to adjust the congestion windows and slow-start windows, smooth the network variation and validated in the system proposed. Experimental result shows that the improved strategy not only improves the video quality in the receiver, but also improves the bandwidth utility.Secondly, due to the scalability of the base layer and enhancement layers of scalable video coding: the base layer uses non-scalable coding to reach the lower bound of the bit-rate range and must be transmitted.The enhancement layer may be truncated into any number of bits or not transmitted. A stream extraction strategy based on the estimated bandwidth is proposed in this thesis, transmitting different spatial layers according to the terminals, and the temporal and quality layers is determined by RTCP receiver report and estimated bandwidth.Finally, this thesis presents the software implementation of the real-time scalable video streaming system on the Darwin Streaming Server and the strategies proposed with experimental results.