An Empirical Model for the Investigation of Effective Intrusion Detection Systems by Using K-Nearest Neighbor (KNN) and Fuzzy (Fuzzy KNN) Algorithms in Mobile Ad-Hoc Network (MANET)

Abstract

A mobile ad hoc network (MANET) is a high-speed network that does not have any infrastructure or centralized management. In recent years, MANET has been popular and widely used by various applications. The important concern of MANET is security. In MANET, Intrusion detection systems (IDSs) are a prominent solution for achieving security. Among these, Clustering-based IDSs are highly noticeable mainly for their proper scalability behaviors. In this paper, the K-nearest neighbor (KNN) algorithm with fuzzy (Fuzzy KNN) inference is proposed to detect the MANET's black hole attack. The implementation of fuzzy inference is effective in the selection of cluster heads. Additionally, Josang mental logic along with beta distribution increases each node’s trustiness. The destination node detects by the trust servers using the reputation and remaining energy. In each cluster, the cluster head is responsible for detecting the node which involved in suspicious activity like black hole attack. The proposed work performance is examined using parameters like total network delay, throughput, packet loss rate, and normalized routing load. The obtained result proves that the proposed system is very effective in detecting the black hole than the other methods.