Research Of Malicious Node Detection For Ad Hoc Networks Based On Trust Evaluation

Wireless Ad Hoc networks are temporary multi-hop networks formed by wireless interconnections without relying on any fixed infrastructures.With the characteristics of rapid deployment,centerless structure and multi-hop routing,wireless Ad Hoc networks have great development prospects.Compared with traditional wired networks,wireless Ad Hoc networks are more vulnerable to malicious nodes.Intrusive malicious nodes attack the network by discarding packets that need to be forwarded intentionally,which degrades network performance seriously.Therefore,the detection and removal of malicious nodes are very important to maintain the network performance.Trust-based detection strategies are common methods for identifying malicious nodes in a network.However,the traditional trust-based detection strategies usually regard the success rate of forwarding packets as an important evidence to detect malicious nodes.The fact that the packet forwarding capability of normal nodes may decrease due to network congestion and time-varying characteristics of wireless channels in Ad Hoc networks is neglected.Thus,the probability of normal nodes being misjudged as malicious nodes is increased.Focusing on the problem that traditional detection strategies are easy to cause false detection,and considering the characteristics of limited resources and changeable channel environment in wireless Ad Hoc networks,this thesis proposes an improved malicious node detection strategy based on trust evaluation.Firstly,the probability of packet loss caused by network congestion is analyzed theoretically,and the results provide a theoretical basis for the design of detection strategy.Then,through innovative improvement on the basis of traditional trust model,the complete design process of detection strategy is given.In the proposed strategy,dynamic trust relationships are established between nodes through distributed autonomous monitoring and interactive collaboration of nodes in the network.A re-evaluation mechanism for nodes with abnormal behavior is designed.Considering the traffic load and forwarding rate of abnormal nodes,further behavioral recognition and trust correction are executed to reduce the probability that normal nodes are misjudged as malicious nodes due to objective factors such as network congestion.A penalty growth mechanism for malicious behavior is designed to improve the ability of detection strategy to resist on-off attack.An adaptive threshold updating mechanism is designed to adjust the decision threshold dynamically according to the changes of network environment,so as to improve the accuracy of detection.The detection strategy proposed in this thesis is modeled and simulated on OPNET platform.The effectiveness of the detection strategy is verified and compared with the other two detection algorithms.The results show that the proposed strategy has higher detection rate and lower false detection rate.Compared with the traditional detection algorithms,it has a certain performance improvement and is more suitable for malicious node detection in wireless Ad Hoc networks.