An Optimization-Based Feature Selection and Hybrid Spiking VGG 16 for Intrusion Detection in the CPS Perception Layer

Cyber-physical systems (CPSs) have become vital to network communication. The CPS combines numerous interconnected computing resources, networking units, and physical processes to monitor the activities of computing devices. The perception layer in the CPS is employed to gather data from the physical surroundings. Still, the interconnection of the physical and cyber worlds creates more security concerns; hence, the operations of the communication networks become more complex. Devices on the CPS perception layer are especially susceptible due to their limited resources. To resolve the issues, the Spiking Visual Geometry Group-16 is developed for intrusion detection in the CPS perception layer. The log file collected from the dataset is normalized using the Quantile Normalization (QN) approach. The major function of QN is to reduce data redundancy. The required features from the normalized data are selected using the Skill Optimization Algorithm (SOA). The proposed Spiking VGG-16 is utilized to detect intrusion. In addition, performance computing metrics like accuracy, precision, recall, F1-score, and Matthew's correlation coefficient (MCC) are utilized for validating the Spiking VGG-16-based model, in which the outcomes of 91.32%, 90.98%, 89.57%, 90.39%, and 90.51% are achieved.