Enhancing cybersecurity: network intrusion detection with hybrid machine learning and deep learning approaches

This study introduces an advanced network intrusion detection system (NIDS) to protect Wi-Fi-based wireless sensor networks (WSNs) using the Aegean Wi-Fi intrusion dataset (AWID). The dataset, which contains multiple classes of attacks, including flooding, injection, and impersonation, is used to train and evaluate the proposed model. The approach employs a robust feature selection process to optimise dataset quality, starting with 130 features, which are narrowed down to 90 relevant ones and further refined to 13 key features critical for detecting security breaches. The data is pre-processed using the standard scaler function, followed by the implementation of a hybrid convolutional neural network (CNN)-based model. The model's performance is compared with other deep learning methods, including deep neural networks (DNN-5, DNN-3) and long-short-term memory (LSTM) networks, using evaluation metrics such as precision, recall, and F1-score. Our CNN model achieves an impressive accuracy of 98% and a low loss of 0.08, with minimal false alarm rates. This research significantly enhances intrusion detection accuracy while reducing false alarms, strengthening the cybersecurity posture of Wi-Fi-supported WSNs in the face of evolving cyber threats.