Toward a Secure Wireless Sensor Network in Intrusion Detection System Utilizing Central-Smoothing Hypergraph Neural Network Optimized With Clouded Leopard Optimization Algorithm

Abstract

Wireless sensor networks (WSNs) is the essential component of wireless technology that provides effective solutions for various monitoring applications. WSN is vulnerable to several security risks as intrusions, attacks, and suspicious activities. Therefore, this paper proposes a Secure WSN in intrusion detection system using Central-Smoothing Hypergraph Neural Network Optimized with Clouded Leopard Optimization Algorithm (WSN-IDS-CSHGNN-CLOA). Here, the input data is taken from WSN-DS database. The gathered data is pre-processed by regularized bias-aware ensemble Kalman filter (RBAEKF) for data cleaning and normalization. The pre-processed data is given into feature selection using Memetic Salp Swarm Optimization Algorithm (MSSOA) to select optimal features. The selected features are given into CSHGNN for classifying the IDS as denial of service (DoS), black hole, gray hole, flooding, and scheduling attacks in WSN. The CLOA is implemented to optimize the hyperparameters of CSHGNN. The performance of the proposed WSN-IDS-CSHGNN-CLOA approach attains 24.39%, 35.71%, and 25.55% higher accuracy; 24.44%, 34.28%, and 14.44% higher precision when compared to the existing techniques.