Towards a Self-Adaptive Super-node P2P Overlay Based on Information Exchange

Abstract

P2P topology design is a hot topic because it is very important for solving the problems such as deficiency in scalability and effectivity of unstructured P2P overlays. A good topology can greatly improve the performance of search algorithm. The paper proposes a self-adaptive Super-node Overlay Based on Information Exchange called SOBIE. The super-node selection in the SOBIE is different from the general super-node selections which only consider physical capabilities such as bandwidth, CPU processing ability, storage space, and etc.. The SOBIE selects the super-nodes by considering the aggregation of the delay, distance, especially the information exchange frequency, exchange time and query similarity. The SOBIE also detects the free-riders and forces them to quit the system. Through experimental simulations, we prove that the SOBIE has better performance than random or standard super-node P2P topologies in terms of file query success rate, the average query hops, and the total number of query messages.