An Energy Efficient Integrated Moving Average (Ima) Model and Hcnd Mechanism with M-Anfis Based Intrusion Detection System in Wsns

Abstract

WSNs (Wireless sensor networks) are distributed, ad hoc networks composed of many tiny, cheap, and resource-constrained sensor nodes with restraints of hardware making it easy to tamper. Yet they are frequently used in hostile locations where they are left unattended, making them susceptible to capture and compromise. Consider the node replication attack, which is exclusive to WSNs and is application-independent. The attacker copies the node after gathering all sensitive data, including the key and secret credentials. Clone nodes are another name for these duplicated nodes. Moreover, multiple operation mode combinations are not considered due to their negligible impacts on power saving or unfeasible operating modes. For instance, putting memories into sleep modes does not save power as CPUs cannot be in sleep modes. For the purpose of detecting replication attacks in WSNs, the current study effort proposes HCND (Hybrid Clone Node Detection) technique using SML-IDS (Supervised Machine Learning Based Intrusion Detection System), an effective technique for balancing energy utilisations with attack detection. In order to detect replica node attacks based on energy, this study effort presents revolutionary IMA (Integrated Moving Average) with HCND (Hybrid Clone Node Detection) technique for detecting replica node attacks based on energy consumption thresholds in WSNs. The network is first divided into parts with inspection for the divided parts. While HCND processes aid in discovering clone nodes in WSNs, Inspection nodes recognize clone nodes by comparing their IDs and keys. This work proposes a modified ANFIS-based IDS with the aim of improving the performances of replica detections. The transitions in IMA model from higher power consuming states (active states) to reduced power consuming phases is controlled by a predefined timetable (sleep and sense states). Future energy losses of sensor nodes are also predicted using statistical measures and when differences between actual and predicted energy consumptions surpass threshold levels, the scheme considers it as malicious behaviour. According on the simulation findings, the suggested detection model has a high detection accuracy and uses less energy.