Research On Malicious Collusion Identification Algorithms Based On Sum Of Squares Of Deviations For P2P Networks

Great changes has been brought to the network world by the rapid development of peer to peer network.A growing number of users and the needs of the growing number of network applications have enhanced network capability.People got convenience from P2P network,but at the same time more and more security challenges also come into the Internet.The malicious nodes give other nodes unfair evaluation and provide false service.Especially,the question of malicious collusion nodes attacking on other nodes group becoming a major issue affecting the network security.To ensure the normal operation of the network and solve problems such as malicious nodes,different nation researchers learn from real-life networks of relationships and establish trust models base on the behaviors of malicious nodes.The models distinguish malicious nodes from others to cut down its impact to network.On the base of reading the relevant literature,this article set forth of models and analyzes the existing trust model which based on different types of theory or mechanism.On the basis of the relevant models which were refined and improved,this paper established Malicious Collusion Identification Algorithms Based on Sum of Squares of Deviations for P2P Networks.Large number of nodes in the network make interactions between nodes and the accumulated the interaction information.Firstly,the local node trust value model consists of evaluation of the interaction between the individual network nodes and nodes.On the base of the local trust model,collecting assessment information for each node in the network to a node,and try to establish a global trust value of node model based on the right to trust value corresponding to different nodes.Combining local,global models have been carried out to establish the difference between the analysis and comparison of the establishment anomaly appraisal recognition model with each of the other nodes in a node to interact with its local trust value global trust value.Taking these three sub-block model,based on sum of squares and theory to evaluate the success node evaluation and failure evaluation as a characteristic value,each cluster node features two levels and the smallest deviation values of two nodes.Clustering the nodes that elevating trust value maliciously we get non-directional malicious collusion nodes set.Clustering the nodes that abating trust value maliciously we get directional malicious collusion nodes set.To achieve the algorithm by flow chart.To achieve each sub-block algorithm process and the total algorithm process.By comparing with the classical algorithm EigenTrust,simulation experiments proved that the algorithm designed by this paper is effective.