The Research Of Intrusion Detection Scheme Based On Weighted Trust Evaluation For Wireless Sensor Networks

WSNs (wireless sensor networks)is a sort of Ad hoc network which consists of large number of self-organized wireless sensors. Due to its characteristic of large scale coverage, randomly deployment, self-organize and environmental adaptation, it is used to collect many kinds of information from the area where the WSNs cover. With the development of computer science, micro-mechanical electronic and wireless communication techniques, the WSNs are more and more popular in many kinds of areas such as military reconnaissance, environment monitoring, traffic control and health surveillance.With the development of wireless sensor network, its security problem becomes more and more serious. Wireless sensors are always randomly deployed in the wild and environment harsh area without any maintenance and protection. So individual sensor nodes are vulnerable to some types of attacks such as packets dropping, fabricating, tampering or physically damage. Once sensor nodes are compromised, malicious nodes can send erroneous data to the base station and possibly mislead monitoring results and influence the whole networks' effectiveness. So it is essential to find new architecture and mechanisms to protect the wireless sensor networks from attacks.In this paper, we first summarize some typical attack models and the corresponding defending and detecting strategies in WSNs. By comparing the strategies with each other to find out their advantages and disadvantages we pointed out the security research problems faced recently and the further researches in WSNs.Second, we develop a scheme for malicious node detection based on weighted-trust evaluation. A weight value is assigned to each sensor node initially. It updates every cycle if the node sends different report from the others. A malicious node will be detected when its weight value is lower than a threshold value. Extensive simulation results show that our improved schemes are more effective in detecting malicious nodes with lower misdetection ratio compared with existing schemes.Third, we apply the weighted-trust detection scheme to military surveillance and reconnaissance application and propose a battlefield information monitoring model. Simulation results show that a more accurate aggregation result can be obtained by comparing nodes' weight sum value and the improved approach is more effective in detecting malicious nodes with lower misdetection ratio in large scale networks even though as much as 25%of the sensor nodes are corrupted.Finally, we summarize the work done by this paper, point out the deficiency and future research of this paper.