CNN-RGU Hyperparameter Tuning for Improving Cybersecurity Intrusion Detection in Industrial IOT Environment

Abstract

Internet of Things (IoT) has revolutionized the manufacturing and industrial sectors by simplifying and making operations more productive. It consists of a networked ecosystem of smart machines, advanced data analytics, and workforce convergence within the workplace. The convergence yields increased production efficiency, improved quality monitoring and control, and improved worker and machine operator safety. However, ensuring security from cyber-attacks and intrusion detection in IoT environment has become a critical concern for modern industries. In recent years, we have witnessed the creation of intrusion detection systems with targeted solutions.  To address these challenges, this paper focuses on hyper parameters' optimization of CNN-GRU using Grey Wolf Optimization (GWO) detecting normal sessions and attack attempts within an IoT environment. Our experimental findings confirm the efficiency of the optimized CNN-GRU model, which outperforms recent comparative studies in performance measures like accuracy, precision, recall, F1-score, and detection cost. Especially, the GWO-optimized GRU model outperforms the CNN-GRU model in multi-class traffic classification with 99% accuracy, 97% F1-score, 98% precision, and 96% recall.