Abstrait

Detection of Intruders in Wireless Sensor Networks Using Anomaly

A.Anbumozhi , K.Muneeswaran

Detecting Intruders in Wireless Sensor Networks plays an important role. Security in network aggregation is not an easy task. Sensor network consist of sensor nodes whose operation can be controlled by underlying network. In this paper, Sensors are used to sense the temperature, humidity, light, voltage etc in a particular area. Extended Kalman Filter (EKF) mechanism is proposed to filter the false data in sensor network. The false data can be acted by some event namely malicious, emergency event. Malicious event are acted by intruders, and Emergency event are acted by some accident occurrence eg. Fire. Intruders make the sensors to get the false reading therefore EKF mechanism is proposed. EKF monitors the behaviour of neighbours and predict their future states, each node aims at setting up normal range of the neighbor’s future transmitted aggregated values. Using different aggregation functions (average, sum, max, and min), theoretical threshold value is calculated. Combining Cumulative Summation (CUSUM) and Generalized Likelihood Ratio (GLR) detection sensitivity can be increased. Intrusion Detection Modules (IDM) and System Monitoring Modules (SMM) work together in order to provide intrusion detection capabilities for WSNs. EKF address various uncertainties in WSNs and create an effective local detection mechanism.