Spiking Neural Network Based Intrusion Detection Framework for Manet Security Using White Shark Optimization

Abstract

Mobile Ad hoc Networks (MANET) are broadly used in crucial application in globally; nevertheless, their dynamic topology, open communication environment and limited resource make highly vulnerable to cyber-attacks, necessitates intelligent and strong security solution. Therefore, in this work, implemented an Optimized Spiking Neural Network (SNN) based Intrusion Detection System (IDS) for improving security of MANET environments. UNSW-NB15 dataset is evaluated, with data preprocessing involves data cleaning and handling missing values for enhancing data quality and feature engineering which includes Exploratory Data Analysis (EDA) and normalization for identifying attack pattern and stabilizing learning feature scale. SNN is designed for modelling the temporal characteristics of network traffic, while the White Shark Optimization (WSO) is employed for automatically tuning the network parameters for enhancing detection accuracy. Using Python software, the proposed work attains an accuracy of 94.71%, precision and specificity of 100%, recall of 94.15%, and F1-score of 96.99%, significantly outperforming conventional detection approaches.