Intrusion Detection in WSN with Optimal Trained Hybrid Model: Introduction to Self-Improved Beluga Whale Optimization

WSN security is a current area of research for many scientists. One of the best security tools for defending the network from malicious assaults or the unauthenticated access is IDS. In this article, we suggest an optimized hybrid classification model-based IDS with certain steps to follow. In preprocessing, the issue of class imbalance data is solved using SMOTE. A collection of features, including raw features, statistical features, and entropy features are extracted in the feature extraction step. Optimal feature selection is done based on the extracted feature set via proposed SI-BWO (Self Improved Beluga Whale Optimization) algorithm. Attack detection is the next stage, in which hybrid classification will be takes place by combining the CNN and DBN model. The same SI-BWO algorithm is used to perform the hybrid model's optimum training. Once an attack has been detected, the attack mitigation procedure is then carried out.