Optimization Method Of P2P Network Traffic Based On Fuzzy Comprehensive Evaluation

P2P(Peer-to-Peer) Network is a logical overlay network built on the top of the physical network with specific characteristics such as network node decentralization, scalability, robustness, high cost performance, load balancing, etc. With wider popularity of the Internet, significant increment in access network bandwidth and the rapid enhancement of computing capacity of terminal systems, P2P network communication has gradually gained popularity and become a hot spot in the field of both application and research for IT industry and academic research. Currently, P2P technology has been widely applied in content distribution, stream media, distributed storage, parallel computing, cloud storage and instant communications. According Internet Survey conducted by CacheLogic, P2P traffic accounts for more than60%of Internet traffic on the backbone of major domestic carriers and it has become the main Internet traffic.However, most P2P applications take random neighbor-selection strategy which ignores the information of the underlay physical network, resulting the mismatch between logical overlay network and underlay network, causing a large amount of redundant traffic which aggravates the traffic pressure of ISPs(Internet Service Provider) and undermines their payment balance. At the same time, the bandwidth crowding situation caused by P2P applications will impose adverse impact on other bandwidth-sensitive applications, such as file downloading and stream media. How to reduce inter-domain traffic redundancy,.ie. P2P traffic optimization, has become a hot issue for both ISPs and P2P content providers. At present, some P2P traffic optimization methods are based on estimated network distance according to single information such as transmission delay, bandwidth, and AS distance.etc. Peers will chose the nearest one as its neighbor with highest priority. However, these methods are sometimes proved to be ineffective in the complex Internet environment.In this paper, a new method named Fuzzy Comprehensive Evaluation is used in neighbor selection to optimize P2P traffic. When choosing neighbors, a P2P peer will take factors such as delay, bandwidth, AS distance, link cost and file difference into consideration to make a comprehensive evaluation of all peers and sort them in desc order. The peer will pick those will higher evaluation result to construct the overlay network. Improved Analytic Hierarchy Process(AHP) combined with Particle Swarm Optimization(PSO) is used to calculate weight of each factor and also to check and correct the consistence of judgment matrix. The results of network simulation experiments show that this method can substantially localize P2P traffic and accelerate file downloading at the same time. This method can further optimize P2P traffic without damaging the performance of applications.