Hybrid Machine Learning Based Black Hole Attack Detection And Prevention Of In Manet

Abstract

A Mobile Ad hoc Network (MANET) is a type of wireless network where mobile devices communicate with each other without relying on any pre-existing infrastructure or cen tralized administration. They are useful in emergency situations but can be vulnerable to attacks. One such attack is the black hole attack, where false route information is sent to intercept communication between nodes, causing major disruptions and rendering the network useless. Hybrid machine learning-based secure AODV (HML-SAODV) is a machine learning-based approach proposed to enhance security in MANETs By modifying the existing AODV routing protocol. This research is organized into phases. The first phase involves gath ering information using the NS-2 simulator, which includes four features related to the destination sequence number. Over 150,000 data were collected and analyzed using both regression and classification machine learning techniques to determine their relationship. The second phase focused on incorporating the suggested approach into the AODV protocol and evaluating its performance in comparison to standard AODV under a black hole attack and other proposed solutions by various researchers. HML-SAODV significantly improves network performance metrics in detecting and pre venting black hole attacks. The packet delivery ratio and throughput are increased by 2.88 and, 2.71 while end-to-end delay and routing overhead are decreased by 2, and 7-fold respectively. The proposed solution does not significantly degrade the network performance when there is no attacker present. A slight increase in the end-to-end delay by 17.4% and routing overhead by 20%. HML-SAODV black hole attack detection rate is above 99%. However, larger networks with more attacker nodes may have a higher false positive rate. Overall, HML-SAODV is a promising solution to enhance wireless ad hoc network security and performance.