Enhanced Negative Selection Algorithm For Malicious Node Detection In Manet

Intrusion detection in MANETS is inevitable as these networks are dynamic, open and no central authority to monitor the nodes' activities. Since the network is open, it allows the node to enter and exit at any point time which provides ease of access to all nodes. Misbehaving nodes utilize these opportunity to enter into the network and cause disruption. This requires building appropriate IDS to monitor the activities of nodes in the network. An Artificial Immune System (AIS) analogous to Human Immune System (HIS) is presented to provide appropriate IDS. The major role of AIS is to classify the samples as self (which are specific to the system) and non-self (which are the foreign body to the system) by means of proposed Enhanced Negative Selection Algorithm (ENSA). The non-self patterns act as a defense mechanism to detect the anomalies caused by invaders. ENSA regards the immune system as a classification system for matching patterns. The proposed ENSA optimizes the detector generation process and performs accurate and precise classification of the network traffic. The ENSA includes two operation detector generation and classification systems. ENSA adapts Particle Swarm Optimization (PSO) technique to enhance the random detector generation to achieve maximum coverage in the non-self space. The classification involves matching the bit strings of the antigen with the generated detectors. The strings which match with the defined detector set are classified as intruders (non-self). The performance result shows that ENSA significantly outperforms other traditional classification algorithms in terms of classification accuracy, detection rate and classification time.