Research On Trust Mechanism In Wireless Sensor Network

Wireless Sensor Networks are often deployed in unattended and hostile environments. Due to the multi-hop communication in WSN(Wireless Sensor Networks), WSN is easy to be attacked in routing and data fusion. Traditional identity trust such as password mechanism, authentication cannot solve attack launched by an internal compromised node. Behavior trust based on historical behavior monitoring and statistical analysis is a necessary complement to reach trusted routing and trusted data fusion in wireless sensor network.To mitigate the selective forwarding attack in routing, data injecting and data tempering attack in data fusion, the dissertation researched a trust mechanism based trusted routing and data fusion. The main works of this dissertation included the following aspects:1. In order to mitigate the selective forwarding attack in wireless sensor network,a behavior trust based reliable routing mechanism is presented. A mixed strategy static game without collusion attack considering is constructed and the payoff matrix of the mixed strategy static Monitor-Forward game, Nash Equilibrium of game and the Quantal Response Equilibrium of the repeated static game are researched. The route trust value of the suspicious nodes can be collected by the sending node through the monitoring strategies which determined by this game.To mitigate the selective forwarding attack in WSN with unreliable wireless communication channel, a continuous strategy Monitor-Forward game is constructed to determine the monitoring strategies of the sending node. In this continuous game,senders choose its strategy from a continuous strategy set, not a strategy set with just two strategies:sleeping and monitoring fixed time. Sending nodes determined its monitoring time by this continuous game, and got a balance between the energy consuming and the false positives on packet dropping detection.The payoff matrix of the continuous strategy static Monitor-Forward game,best reaction function,Nash Equilibrium of game and the Quantal Response Equilibrium of the repeated static continuous game are researched.2. To mitigate the collusion selective forwarding attack in WSN, a two-stages dynamic Monitor-Forward game is constructed in the suspicious paths whose packet loss probability is greater than a threshold to determine the monitoring strategies of the sending node and collect the route trust value of the nodes in these paths. Awake-up mechanism is applied in the second stage of the dynamic game to save battery energy of the sender node and detect the collusion attack..The sending will determine whether wake up its another neighboring node to monitor the The payoff matrix, Nash Equilibrium and the repeated Quantal Response Equilibrium of this dynamic game. The dynamic Monitor-Forward game is researched and simulated in different game parameters.3. In the wireless sensor network with Cluster-tree structure, a variable weight fuzzy comprehensive evaluation based trust value computing is researched to improve the accuracy and reliability of the secondary data fusion and reduce fusion weights of cluster head nodes with some defect. Time delay, data amount and trustworthiness of each cluster head will all affect the final fusion weight.However, the influence of these three factors is not fixed. By increasing the weight of the factor with too low value in the variable weight based fuzzy comprehensive evaluation, the clusters with some deficiency will be given prominence and therefore, the cluster head node with deficiency will get a small fusion weight and have a small influence in second data fusion.4. In order to store and query the data reliability reputation of cluster head nodes and avoid the bottleneck of centralized storage securely, a distributed hash table based Divisional Dual-Chord based Cluster Head Reputation Storing and Querying Mechanism(DDChord) is proposed and researched. The data reliability reputation of ordinary nodes in a cluster are centralized stored in the cluster head. By dividing the whole wireless sensor network into several partition, and distributing a node’s reputation in the partition it was located, the more hop to transmitting the reputation between the storing node and the aggregation node can be avoided. The simulation results show that if the number of nodes in WSN is large, the query between partitions are much less than query inner a partition and DDChord can get shorter average query time than traditional Chord protocol.By applying the behavior trust mechanism in wireless sensor network routing selection and data fusion, the selective forwarding attacks launched by the compromised nodes with legal identity will be mitigated, and a more reliable fusion results can be obtained. Trust mechanism is a necessary complement to existing identity trust in wireless sensor networks.