Research Of Routing Algorithm In Unstructured Peer-to-Peer Network For Heterogeneous Environment

This paper is supported by the National Natural Science Foundation of China under Grant No.90204010.In the research of Peer-to-Peer network, the question of heterogeneous may lower the efficiency of routing algorithm and increase the overhead of the whole system in some extent. This issue includes geographic heterogeneity, host capacity heterogeneity and session duration heterogeneity in different degrees. We study on the heterogeneity of P2P networks in this dissertation and make the following contributions.We study the host capacity heterogeneity. All of the algorithms assume that all nodes have the same capacity to process message. However, the capacity of the nodes in the P2P network is extremely heterogeneous, which leads to load imbalance. Aimed at improving the degree of load-balancing, we proposed the virtual-node algorithms to balance the workload, which divided and combine the nodes into same unit. Virtual-node algorithm is able to balance the load efficiently.We study the geographic heterogeneity. Aimed at improving the efficiency of P2P routing algorithms, we proposed the delay-aware algorithms based on different aggregation unit. The overhead of the algorithms is different and so is the efficiency improvement.We study the heterogeneity of session duration. P2P network have the limitation of high maintenance overhead and poor adaptability to highly dynamic network environment. This paper proposed session aware algorithms which based the availability of file. This algorithm evaluated the failure rate of the node. Experimental results show that the algorithm is effective in reducing the maintenance overhead and improving the availability of P2P resource greatly.