Detection of Intrusion Data using Deep Learning for Multi Class and Binary Class Data

In today's world, Wi-Fi networks are almost everywhere connecting our phones, laptops and smart devices, which make our life easier but also exposes to a variety of cyber threats. Malicious users can exploit vulnerabilities in Wi-Fi systems to perform attacks such as flooding networks, impersonating legitimate users, or injecting harmful data packets. Traditional intrusion detection system are often unable to attack these attacks efficiently and tend to generate many false alarms, which reduces trust in the system. This research proposes a new approach that combines artificial neurons with a bioinspired optimization technique called Harris Hawk optimization (HHO). We used the AWID dataset to train and test the model carefully preprocessing the data and selecting the important features. The result shows that the proposed system can detect attack with an accuracy of 99.16%, a detection rate of 99.49, ana a very low false alarm rate of only 1.27%. Compared to other machine learning and optimization methods, this model performs better, making it a practical and effective solution for Wi-Fi security.