Malicious Node Detection in Wireless Sensor Networks Using Support Vector Machine

Wireless Sensor Networks are basically a collection of sensor nodes scattered in a large area such that the desired information can be collected. But sensor nodes are also vulnerable to the attacks such as malware, hackers, faulty hardware or from the physical phenomenon etc. Hence, it is mandatory to protect a sensor node from an attack because if it gets attacked then the information sent by the sensor could be wrong and lead to incorrect data analysis and hence can lead to unnecessary outcomes. In this paper, a technique based on machine learning to discover malicious nodes within a randomly deployed sensor network has been proposed. Here, Support Vector Machine for time series prediction has been employed to find the malicious nodes based on the past values acquired by those individual nodes.