Study On Resource Sharing In Unstructured P2P Networks Based On Sociological Characteristics

As the widespread popularity of the Internet and the rapid increase of the bandwidth, file sharing based on Peer-to-Peer (P2P) networks has occupied an absolute share in the network flow and thus became the major mode of Internet file delivery. As a major design pattern for future systems opposite to the traditional client-server paradigm, research on P2P networks is extremely important and can significantly change our day-to-day use of Internet. A study shows that the recent large-scale Internet traffic was generated by P2P file sharing system. However, there are still a lot of fundamental problems in the P2P file sharing networks:1. There exist numerous malicious nodes in unstructured peer-to-peer networks, whose fraud behaviors decrease the efficiency and reliability of the resource search.2. Due to the application of excessive users, the hot spot will be overloaded, thus fail to provide its service.3. Replica consistency maintenance can not cover both file update efficiency and its coverage at the same time.This paper chooses the unstructured P2P network as the research object, making a deep study on the similarities and differences between social network and unstructured P2P network. With above three problems as the three basic points, the paper is focused on the sociological characteristics in the unstructured P2P network. 1. Propose a resource discovery approach based on the trust and social networks (TSN). In unstructured peer-to-peer networks, there exist numerous malicious nodes, whose false behaviors decrease the efficiency and reliability of the search. Therefore, this paper is devoted to decrease the negative effects of those nodes in resource searching process by regarding P2P networks as social-like networks and discovering resources based on its small world characteristic and trust relationships. The new concepts of forwarding trust, routing trust and transaction trust that evaluated according to previous behaviors and transactions history of the node with other nodes are proposed to assess the a node's ability of providing reliable services. When forwarding the query message, a peer node will give priority to the friend node with high trust and correlation. The simulation experiment results show that TSN can effectively screen out the malicious nodes and at the same time ensure the reinforce of the honest peers'cooperation and their remarkable improvement of the query efficiency2. Propose an adaptive file replication method based on dynamic community (ARDC). To solve the problems of inability of placing the replica in the area with large demand and the blindness of the replica location in existing the hot spot reduction strategy in current unstructured P2P network, due to the inconsideration of applying the information search history and thus it neglects the ubiquitous relation among the nodes, this paper presents a virtual adaptive organizational community method addressing the issue of where replicas should be placed with the reference of human social community and sociological characteristics. When the node is under load, the accidents of hot spot will be reduced by adaptive proactive replication of the replicas to the static community applying the locality and popularity of the file. When overloaded, the frequently-queried node will adaptively build the dynamic community, therefore the amount of the replicas decreases and the discovery rate of the replicas increases. Simulation results demonstrate the efficiency and effectiveness of ARDC, compared with other approaches. It dramatically improves the replica hit rate, shortens the average path length and reduces the overhead of file replication, which means it reaches relatively high query efficiency with comparatively low replication overhead.3. Propose an adaptive file consistency maintenance method based on peer-to-peer community (CMC). Most consistency maintenance methods propagate update messages based on message spreading without considering building a community-based dynamic replica maintenance network in the application layer utilizing the network topology characteristic, failing to keep the balance between the reduction of the excessive file update information and the consistency maintenance of all the replicas. This paper proposes to build replica consistency maintenance network with the sociological characteristics of power-law distribution and small world, using the nodes with high utilization degree as the key node to forward the update information. When delivering the update information, the owner node will choose the nodes and delivering path according to the topology-structured replica maintenance network. At the same time, when missing the update information, the replica node will require the consistency update information through its adaptive polling mechanism. Simulation results demonstrate the effectiveness of CMC in comparison with other approaches. It can dramatically reduce overhead and yield significant improvement in the efficiency of consistency maintenance approaches.