ML Based Intrusion Detection Scheme for various types of attacks in a WSN using C4.5 and CART classifiers

Abstract A key component for preserving a WSN’s integrity is designing an intrusion detection system (IDS). This article covers several sorts of security threats that may occur in a WSN. It proposes a detection of malicious nodes in a WSN. This system identifies four common assaults (i.e., black hole, wormhole, gray hole, and DDoS attacks) and is based on a Base Station machine learning (ML) algorithm (BS). The suggested ML algorithm continually analyzes data patterns from each node. Based on this technique, BS recognizes the network’s harmful behavior and provides notifications to neighbor nodes to prevent attacker. First, the various assaults are examined and their characteristics are derived in terms of network parameters. Furthermore, the gathered pattern trains an ML algorithm. Then, the attacker node is effectively and properly classified within the BS. The NS2 simulator simulated the needed secure WSN. The experimental findings revealed excellent accuracy of the suggested method for attack detection. This precision can enhance network performance in terms of energy usage and packet delivery (PDR).