Research On Pervasive Service Discovery Strategy Based On Model Of Peer To Peer

With the era of pervasive computing comes, variety of service discovery technology that support pervasive computing environment is researching. Considering the high degree self-organization of pervasive computing environment.P2P service discovery algorithm provides a good theoretical basis for researching service discovery in pervasive environment.At present, the P2P resource discovery algorithms are concentrated indexing algorithms, structured algorithms, unstructured algorithms and mixed-discovery algorithm. Napster system as the representative of concentrated indexing algorithm, using a centralized directory server mechanism, there is single point of failure problem in this algorithm, the directory server will be the bottleneck of the entire P2P system, once the central server have any problems, will cause the entire system collapsed. Gnutella is the representation of unstructured algorithms, it using flood or similar flooding algorithm, each user broadcasts the message to the user which directly linked to it, after these users received the message, they also broadcast the message to the respective users who connect it.By analogy, until the request is answered or the TTL value of a message reduce to O.The reliability of the algorithm is poor,and the consumption of the network resource are huge.The typical representative of structure algorithms including Tapestry, Pastry, CAN and Chord, all of them are using a distributed hash table (DHT) data structure,and according to different algorithms they decide the way of the nodes maintain hash table in network. In these typical representation of structure algorithms, Chord algorithm has the best performance and used the most widely. Mixed-discovery algorithm combines concentrated indexing algorithm and fully distributed algorithm to construct the network topology.In this paper, we focus on how to enhance the discovery efficiency and coverage of the structure algorithm. Based on the Chord algorithm, combining with Small World theory, we put forward a new algorithm for fingertable construction. The traditional DHT discovery algorithm each node maintain the fingertable that store node information of adjacent node, In order to build Small-World Model, This issue put forward the idea to add a remote node information. The fingertable of each node delete redundant information by calculating, add the corresponding remote index.It is different from algorithm of select the remote connection node randomly. This paper selects the remote connection node by calculateing local node, after adding a remote connection node it can assure both make the range of service discovery cover all network and not increase the length of fingertable, simplifies the calculation of the fingertable and maintenance work. The simulation proved that the algorithm can reduce the path length of service discovery effectively, improve success rate of service discovery.This paper formed by the five chapters. The first chapter describes the resource background of discovery research and research status of P2P model; The second chapter describes the characteristics of P2P networks and search models of P2P network, these theory laid the foundation of service discovery in pervasive environment that we research; The third chapter describes the traditional Chord and Chord algorithm, and analyse the shortage of them; The fourth chapter gives an improved method of Chord algorithm. And theoretical analysis of the improved algorithm, simulation and simulation results analysis; The fifth chapter is the summary and prospects.