TS-GCRA: trend factor smoothing greater cane rat algorithm with Attention-Based Autoencoder Belief Network for robust intrusion detection in wireless body area network

WBAN have several medical sensors with different functionalities and data rates, allowing for continuous patient health monitoring and fast response in emergency situations. Intrusion Detection Systems (IDSs) are usually used as a second level of protection to scan nodes and the network for any criminal activity. But the research community has not very much focused on intrusion detection technologies designed specifically to address the special needs of WBANs. In this research, an advanced IDS by employing optimization based deep earning approach is developed. Initially, WBAN simulation is performed, followed by data preprocessing, which includes data cleaning and data normalization, is executed. Then, significant features from data are selected with a novel optimization algorithm, called proposed Trend Factor Smoothing Greater Cane Rat Algorithm (TS-GCRA), which merges Trend Factor Smoothing with Greater Cane Rat Algorithm (GCRA), seeking to minimize both time complexity and memory usage. Intrusion detection is finally performed using the proposed Attention-based Autoencoder Belief Network, which combines Attention-based Autoencoder with Deep Belief Network (DBN). Subsequently, the suggested model efficiently identifies intrusions in the WBAN system. Experimental analysis stated that the proposed approach reached an accuracy of 96.8%, a precision of 78.5%, recall of 97.4%, and an F1 score of 86.1%.