Research On Non-cooperative Behavior In Clustered Routing Of Wireless Sensor Networks

In self-organizing wireless sensor networks,all nodes,existing equally and independently without relying on any routing infrastructure,complete information transfer by relaying data with one another.So,only with the cooperation of all the participating nodes,could self-organizing wireless sensor network operates normally.How to deal with the non-cooperative nodes is the focus of the study.Non-cooperative nodes can be classified into two types,the intruded malicious nodes and the resource-saving selfish nodes,both of which are with legal identities.However,malicious nodes could attach and damage the network actively,while selfish nodes tend to enjoy network service but are unwilling to undertake the cost.Both types of nodes can affect the network performance.To solve the above problem,this thesis carries out study on non-cooperative behaviors of wireless network and puts forward the methods to detect malicious and selfish nodes and to strengthen cooperation.In summary,the major research contents of the study include mainly the following two aspects:1.To detect internal malicious sensor nodes,a double cluster heads based feedback trust model(DCFT)is proposed that can guarantee data reliability and integrity during transmitting,sensing and aggregating processes.Immediate interaction results are used to assess the direct trust of sensor nodes including communication,data sensing and aggregation trusts.Communication trust between neighbors are evaluated by Bayes formula.The cluster heads independently assess data perceived trusts for their member nodes according to the spatial correlation of local data.Interactive monitoring mechanism and trust feedback method are introduced that temporal correlation is used by base station to assess data fusion trust for cluster heads.Moreover,to enhance the objectivity and accuracy of trust assessment,historical trust is introduced by a time window and used as the supplement of the direct trust.The weights between direct and historical trusts are designed dynamically according to the frequency of immediate interaction.The simulation results show that the proposed method can effectively detect data fault,data exception and internal malicious or compromised nodes.2.Sensor nodes in Wireless Sensor Networks(WSN)are rational insufficiently for decision making under conflict environment.Recently,game theory is introduced into WSN to solve the clustering problem for selfish sensor nodes.However,lots of classical game theory based clustering protocols are not practical for resource-limited nodes.This is because game players need to know the information of the whole network,and sensor nodes are assumed rational enough in these protocols.In this thesis,an energy-efficient clustering routing algorithm based on evolutionary game(EECEG)is presented for WSN.The replicator dynamics equation is introduced to prove the existence of the Evolutionary Stable Strategy(ESS).Sensor nodes are modeled as selfish game players which have the choices of declaring to be a CH candidate(D)and declaring not to be a CH candidate(ND).By observing and simulating the behaviors of rivals,each player makes decision based on the factors of residual energy and numbers of neighbor nodes repeatedly until arriving at the equilibrium for payoff.The experimental results indicate that the protocol EECEG can prolong the network lifetime and balance the energy consumption among sensor nodes.Moreover,energy efficiency is highly enhanced for data transmission.