DETECTION AND PREVENTION DOS ATTACKS IN MANET USING INTRUSION DETECTION SYSTEM AND SUPPORT VECTOR MACHINE (SVM)

Abstract

The Mobile Ad hoc Network (MANET) is a collection of moveable wireless nodes. These wireless nodes relate to wireless links i.e. radio waves. MANET nodes have full freedom such as open medium for communication, dynamic topology, and without any central control. MANET possesses popularity in various applications such as military, rescue, earthquake, and disaster operations. However, the mobility and freedom of Mobile nodes creates vulnerabilities to various routing and DoS attacks in MANET. Moreover, the nodes are prone to various attacks where a malicious node drops the data packets during communication and thus, reduces the network performance. To solve the above-mentioned concerns, a malicious node detection method is proposed using Support Vector Machine (SVM)-a supervised learning approach and intrusion detection system (IDS). In this paper, we proposed a hybridized technique (IDS-SVM) for identification of the malicious nodes. In the proposed technique, the IDS node searches the data items and retrieves closest neighbor nodes within the network range using Euclidean distance. IDS nodes are considered as query points and forwards the status packet periodically to judge the behavior of other nodes. IDS-SVM is implemented with the cluster approach to avoid beacon messages to route overhead in the network. Meanwhile, intermediate nodes are associated with the IDS nodes in a specific range. Simulation results indicate that IDS-SVM achieves consistent overhead routing and network delay. Our findings could also indicate that the proposed technique (IDS-SVM) has obtained a high accuracy rate in the detection of malicious nodes, thus it is fast and efficient from the perspective of MANETs.