Machine Learning-Based Detection And Prevention Of Malicious Node Behavior, Fraggle, And Advanced Smurf Attacks In Wireless Sensor Network

Abstract

This research focuses on enhancing the security of Wireless Sensor Networks (WSNs) against harmful actions by using a combination of simulations and advanced machine learning. A custom network environment is created in a simulator, containing 30 nodes with 3 malicious ones. This setup imitates real situations, letting us watch how the network behaves under different conditions. Data is collected and analyzed, capturing important factors like power use, delays, and how packets act. These factors are used to teach machine learning models like Support Vector Machines, Naive Bayes, Logistic Regression, and k-nearest Neighbours. These models can tell apart normal and malicious behaviors effectively. To stop harmful data from entering the network, a decision tree model checks incoming packets. If they're labeled as malicious, they're blocked. This system works continuously, adapting to new dangers. The research offers a full security approach, protecting WSNs from various attacks. By mixing simulations, analysis, and machine learning, it highlights the power of using different fields to make networks stronger. The results guide us toward secure and adaptable WSNs that can handle security challenges in changing situations.