Research On P2P Resource Sharing System Based On Wireless Mesh Networks

P2P (Peer-to-Peer) technology changes centralized and client-server (C/S) models by making peers share information and computing resources directly in a virtual overlay network without servers. Currently, resource sharing is the most typical and popular P2P application. In wired networks, researchers focused on the topology structure, search algorithms of P2P resource sharing systems, and have seen some gratifying accomplishments. With the development of the wireless network technology and the widespread use of mobile terminals, most researchers have proposed many approaches to realize wireless P2P resource sharing in Mobile Ad-hoc Networks (MANETs).Wireless mesh network (WMN), a relatively static multi-hop wireless network which is extended from mobile ad-hoc network (MANET), is an emerging technology for addressing the provision of high bandwidth to users at low cost, anytime, anywhere, and enabling the creation of new types of applications and services for clients accessing the network. Some researchers began to think to apply P2P technology in WMNs. The article aims at a P2P resource sharing system in WMN. Analyzing the critical steps of realizing P2P resource sharing, we propose the structural model of the system, which is closely related to the features of WMN, then discuss the problems of enabling P2P file sharing in WMN. The main work and innovation points the paper as followed:The second chapter proposes the structural model of the P2P resource sharing system in WMN. According to the features of the WMN, a two-tier architecture is put forward. The topology of overlay network, different functions and duties of mesh clients and mesh routers, and the management style of shared resources are prescribed. In the two-layer P2P resource sharing system, the mobility of mesh clients, which are the shared resources providers, causes the performance degradation of resources sharing. By borrowing from the management style of shard resources, a location management mechanism for mobile mesh clients is proposed. A manager router which assigned to each mobile client takes the responsibility of peer location. The simulation results show client location management mechanism is effective in decreasing message overhead and response delay in P2P resources sharing system with high churn rates.The third chapter proposes a new algorithm to construct the topology of the overlay network based on Chord——SpiralChord. SpiralChord proposes an ID assignment technique based on spiral curve which uses location awareness to resolve the problem of topology mismatch and cross-layering to realize the search algorithm based on1-hop broadcast communication. According to studying the requirements of location awareness and cross-layering to the construction of overlay network topology, a curve selection rule based on features of space filling curves and minimum average of the sum of value differences is proposed. After analyzing the several space filling curves, the spiral curve which satisfies the curve selection rule is used to be the basis of constructing the overlay network. According to the structure of spiral curve, a generation algorithm of spiral curve is proposed to assign IDs to mesh routers, which construct a ring overlay network. Simulations show SpiralChord is more effective in reducing message overhead and increasing lookup performance than Chord.The fourth chapter proposes a weight-based multi-keywords information retrieval mechanism in SpiralChord model. We use the method extracting and restructuring the hash values of N keywords with the highest weight to get a new hash value which represents the resource. By allocating IDs to each resource based on hash values, a P2P peer can store or search the information of resources related to multiple keywords. The information retrieval mechanism describes a resource by using N keywords with the highest weight. The similarity of resources can be judged by their hash values, the smaller the differential, the more similar they are. By dividing the resources into clusters on the basis of similarity, the resources in the same cluster are allocated IDs according to the spiral curve and the ID assignment of routers, which making the information of resources in the same cluster being stored in routers with close-by locations to realize the cluster of similar resources. Simulations show the algorithm is more effective in reducing message overhead and increasing lookup performance and precision ratio, supporting fuzzy retrieval. The fifth chapter proposes a resource provider selecting method based on physical programming. In resource sharing system, the number of the copies of a resource is not only one. Good resource provider can assure better transmission efficiency, decreasing the system consumption. After analysing and discussing the factors with influence on the communication and data exchange between peers in wireless network, we choose appropriate evaluating indicators of mesh clients, propose a resource provider selecting strategy based on physical programming. The strategy assess the resources holders according to their values of aggregate preference functions, satisifing the one vs others criteria rule. In order to reflect the personalization of resource provider selection, the analytical hierarchy process is used to calculate the preference coefficients of evaluating indicators, the requesting peers can choose different resource providers according to different preference strategy. Simulations show the algorithm is more effective in increasing the successful ratio of resource transmissions, reducing the average download time.