Research On Key Technologies Of Trust Evaluation In Wireless Sensor Networks

With the development of Wireless Sensor Networks(WSNs)technology,WSNs have been used in many aspects of human life,such as telemedicine monitoring,intelligent transportation,environmental monitoring and so on.Nodes in WSNs are usually deployed randomly on an unattended monitoring area.The duties of sensor nodes are monitoring events occurring in the area and sending the information to the base station.How to ensure the security of sensor nodes is one of the research focuses of WSNs.Traditional cryptographic methods can effectively defend against external attacks,but internal attacks are defended primarily by evaluating the trust values of sensor nodes.Existing WSNs trust models mainly evaluate trust value according to the interaction behaviors between nodes,so malicious behaviors will reduce trust value.Some malicious behaviors are easy to monitor,such as packet tampering attack,black hole attack and selective forwarding attack,however others are difficult to monitor.It results WSNs trust evaluation methods are less resistant to other attacks.In this dissertation,according to different attacks,we optimized deployment strategies for WSNs,improved trust calculation process,and proposed trust evaluation methods for improving network performance based on the research of existing trust models.The main achievements in this dissertation are as follows:(1)For malicious behaviors such as packet tampering attack,black hole attack and selective forwarding attack which attack packets directly,this dissertation proposes a coverage balancing based trust evaluation method for WSNs.The sleep mechanism is used to optimize network topology and ensure the uniform distribution and the coverage density of nodes.In addition,this dissertation proposes a method for reducing energy holes for WSNs trust evaluation,which ensures the network be always fully connected.Simulation experiments show that this method can improve the accuracy of trust assessment.In addition,it reduces the nodes overhead,and effectively improves the average lifetime of nodes.Compared with existing trust models,this method has higher success rate of event transmission.(2)For packet replay and flood attack that consume node energy,this dissertation proposes a trust detection method for WSNs based on artificial immunity.This method introduces artificial immunity into trust detection of WSNs.The three trust attributes(communication trust,energy trust,and data trust)are used as a trust vector,and the negative selection algorithm in artificial immune is used to match nodes trust vectors.Simulation experiments show that this method can effectively defend against packet replay and flood attack,and can quickly detect malicious nodes and reduce the damage to the network.(3)For direct attacks,we choose two typical attacks,which are selective forwarding attack and flood attack.This dissertation proposes a trust evaluation method for WSNs based on information entropy.Each round,nodes will first evaluate the information entropy of refusing forwarding and the information entropy of sending actively with neighbor nodes,and judge whether neighbor nodes have selective forwarding attack or flood attack.Then nodes determine whether sending trust value of neighbors to the base station or not.This method reduces trust evaluation of global nodes to trust evaluation of partial nodes.Simulation experiments show that this method can effectively defend against selective forwarding attack and flood attack.(4)For sinkhole attack and wormhole attack,which are indirect attacks,this dissertation proposes a grid trust evaluation method for WSNs based on similarity measure theory.It reduces nodes overhead by evaluating trust value of grids instead of trust value of nodes.This method introduces the similarity measure theory of fuzzy mathematics,and regards the interactions between grids and the grids trust value at nodes as fuzzy sets.First,it evaluates the similarity measure of interaction to find suspicious grids,and then evaluates the similarity measure of trust to distinguish malicious and legitimate nodes.Simulation experiments show that this method can effectively defend against sinkhole attack and wormhole attack.