Research On Performance Optimization Of Network Coding-based P2P Media Streaming

P2P media streaming can utilize the host nodes'bandwidth, computing, storage and other resources, so that node can not only receive streaming data from other nodes, but also provide data services for them. It has a good scalability and cost-effective which can greatly reduce the bandwidth consumption on the streaming server and avoid the local network's congestion. Network coding has emerged as a promising technique to break through the restrictions that intermediate nodes can only perform data storage and forwarding operations in the traditional means of communication, in which intermediate nodes are allowed to encode incoming data blocks to produce coded outgoing ones, and the original data blocks are decoded at receivers. A lot of theoretical results show that network coding can improve the whole network's throughput and reduce the impact of network dynamic nature to the system.In view of the potential advantages of network coding, scholars at home and abroad begin to research network coding-based P2P media streaming, and have achieved great results. But on the whole the study of network coding-based P2P media streaming is still in its initial stage, facing many challenges, including how much impact does network coding have on the system's correlative performance indicators, how to construct overlay topology so as to take full advantage of network coding, how to make data scheduling optimization considering the coding data's "hybrid" characteristics, and how to optimize the whole system performance considering both topology construction and data scheduling and so on. Around these challenges, the thesis studies several key issues and proposes effective solutions.(1) Topology optimization of network coding-based P2P media streaming.The existing topology construction method of network coding-based P2P media streaming still adopts the approach based on the neighbor nodes'random selection, according to some examples, the thesis shows that this method doesn't fully consider the characteristics of network coding, leading to linearly dependent blocks that are impossible to be used to recover original blocks and consume bandwidth. The thesis studies the reason of linearly dependent problem, analyses the relation between topology and this problem, and presents the topology construction rule that can effectively avoid this problem.(2) Data scheduling optimization of network coding-based P2P media streaming.In the study of network coding-based P2P media streaming, pure pull or push strategy are commonly used in the process of data scheduling. Pull strategy has a great data scheduling delays and control overhead, while push strategy will make nodes receive several redundant data blocks. In order to overcome the lack of pure pull and push strategies, based on push strategy the thesis analyses the reason of data blocks redundancy and proposes a new scheme for P2P live media streaming based on network coding called customR2. Using push-pull strategy this scheme greatly diminishes the redundant blocks and improves the nodes'playback quality. Simulation results demonstrate the effectiveness of the proposed scheme.Based on the research results of network coding theory, the thesis studies the performance optimization of network coding-based P2P media streaming, providing a good theoretical and technology foundation for deploying network coding-based P2P media streaming systems.