Research On Resource Searching In An Unstructured P2P Network Based On Markov Chain

Recently, peer to peer network technology has become a hot issue due to its widely application in resource sharing, cooperative work, real-time communication and many other fields. Among them, the instructed P2P network is widely used with the advantages as following. It has simple topology. Increasing and decreasing nodes have small effect in network system. The network supports obscure inquiry. The search mechanism is easy to achieve. Existing unstructured P2P resources searching algorithm is mainly based on the modified flooding algorithm, and this kind of algorithm has two aspects of problems.(1) The efficiency of resources searching is low due to being blind search.(2) The algorithms produce a large number of queries redundancy package, so that it caused the network load. Aiming at the shortcomings of the flooding algorithm, the mainly researching work of this thesis embodied in the following two aspects.This paper proposes resource searching based on interest factor algorithm. The Markov Chain model is applied to the path selecting in the process of resource searching. The proposed algorithm introduces the concept of node interest factor in transmitting query packet, and node interest factor is used to determine the state transition probability. On the basis of this, the algorithm chooses the node that most likely to send back the query feedback package as the next hop. The algorithm could guarantee that the random sampling procedure converge to the extreme value with the fastest speed. According to the simulation, the query efficiency of the algorithm is superior to the traditional flooding algorithm, and the count of the search hop of the algorithm is smaller.The resource searching algorithm based on interest factor only consider the interest of node, instead of taking the load balancing problem of the network into account. This paper proposes a resource searching algorithm based on load factor. This algorithm take load factor and node interest factor into consideration, and construct matrices based on these two factors, so that resource searching can expand sampling along the direction of the factor’s expansions. Experiment shows that, this algorithm has higher search efficiency and a more balanced network load balance.