Intrusion Detection in Wireless Sensor Networks

Wireless Sensor Networks are prone to many effects due to its openness characteristic. In this Paper we are going to inject a selfish attack to some nodes which would act selfish and would not send packets to destination. There are two types of nodes, Sensing node and processing node. The sensing node will sense the action of every node. The processing node will participate in the routing activity to forward the packets. The processing node either drops the packet or sends the packet back It stops normal behavior of the packet route. In order to find the malicious node a threshold limit is maintained for every node, if it exceeds the limit the particular node is malicious node. Whenever the limit is exceeded an alarm is raised. All alarms raised are handled by sensing node. The sensing node will filter the alarms that are valid and generate an alarms file with Node ID and alarm threshold value. The alarm file is given to Naive Bayes algorithm which will generate the True positive and True negative rate of the malicious and non-malicious nodes.