Research On Fault-tolerant Mechanism Of P2P Media Streaming Network

P2P media streaming service is one of the most widely-used services in Internet. It attracts a large number of users from all over the world. P2P traffic represents a big share of the whole Internet traffic. P2P media streaming network is constructed in the unreliable transmission links and the behaviors of its users is unpredictable. There are some factors will affect the Quality of Service (QoS) of the whole P2P media streaming network when a media content is streaming, such as frequently online and offline behaviors of users, network congestion, network heterogeneity, instability service capacity of users and so on. So, fault-tolerant mechanisms must be designed to ensure high quality of service and provide a satisfied Quality of Experience (QoE) for users.Firstly, this thesis studies and summaries three kinds of existing fault-tolerant mechanisms in current P2P media streaming network, they are: 1) Encoding based fault-tolerant mechanism, which is deal with the packets loss in transmission; 2) Service migration based fault-tolerant mechanism, which mainly handle the problems such as node failure, link failure and service capacity decreasing. 3) Redundant data based fault-tolerant mechanism, which distributes redundant data among nodes by using replication to improve performance of fault tolerant.Secondly, this thesis analyzes small-world theory and small-world network models. A high clustering coefficient and low average shortest path length feature of small world can improve robustness and throughout of network and facilitates efficient search. Applications of small-world theory in P2P network are studied in this thesis, especially a small-world based P2P media streaming model BEAM (Bit strEAMing). On the basis of principal of BEAM, a new fault-tolerant mechanism is proposed which is named as SWFTM (Small-World Fault-Tolerant Model). SWFTM organize ordinary nodes into clusters by super nodes. Forming cluster scheme, Stabling cluster scheme and Leaving cluster scheme are designed respectively.In simulation experiments, SWFTM compares with BEAM in terms of forming cluster time, checking node failure message cost and network stability when lots of nodes failed simultaneously. The results show that SWFTM performs better than BEAM. In conclusion, SWFTM is very suitable for P2P media streaming network as a fault-tolerant mechanism and can cope with node failure.