Improving Wireless Network Using Hyper Tuned Extra Tree Classifier as Intrusion Detection System

For past decade, one of the evolving and critical modes to connect the individuals and the devices over the worldwide is wireless communication. With the exponential growth of internet usage, the prevalence of cyber-attacks has surged, necessitating robust intrusion detection systems (IDS) for network security. The increase in the degree of sophistication and complexity over cyber assaults necessitates a proliferation of effective and intelligent security methods for shielding the infrastructure of the network. This study presents a novel supervised Machine Learning (ML) approach aimed at enhancing network security through accurate classification of network traffic as malicious or benign. The major application of ML in wireless communication is dynamic spectrum management. When the radio spectrum grows increasingly crowded, effective spectrum usage is important. Employing a blend of supervised learning algorithms and feature selection techniques, the model maximizes detection success rates by identifying pertinent features and leveraging advanced algorithms in NSL-KDD dataset that consists of target attack types identified by AODV routing protocol. Moreover, the performance of ML methods has improved further by hyper parameter tuned and it is measured using metrics of confusion matrix. Randomized search based hyper parameter tuned Extra Tree Classifier (ETC) improves accuracy to 98.90% when utilizing hyper parameter tuning to determine optimal IDS in the wireless network.