Malicious node detection based on clustering techniques in network

Abstract Malicious node is an exciting area of research which has been impacting negatively over the performance parameters of the network. The attacker node can affect the throughput of the network causing the degradation of the network response time. To deal with it, anomaly based intrusion detection techniques have been used. Behavior based detection observe the behavior of the existing node by analysing its parameters and determine the type of the node. Intelligent mechanisms in the form of machine learning are developing newer techniques for localisation and removal of such malicious nodes from the network. The main aim of the study is to analyse the features of the various networks and identify the category of the node based on some particular characteristics by using unsupervised machine learning. For determining the malicious node, clustering techniques are tested on the reduced CIDDS-001 dataset and it was found that the density based clustering algorithm is performing best in identification of malicious nodes in the network.