Deep Variational Auto Encoders for Botnet Attack Detection in Industrial Internet of Things
K A Sharada, Chahat Gulati, B Subha, Md. Abul Ala Walid, Manasi Vyankatesh Ghamande, Neerav Nishant · 2023
With the rapid increase of the adoption of Industrial Internet of Things (IIoT) in manufacturing industries, the frequency of cyber-attacks like botnet attacks on IIoT devices is also reaching historic highs making it predominant to swiftly recognize these attacks to lower the subsequent potential risks. It is essential to formulate a botnet attack detection method to accelerate instantaneous alerts to interrupt the connection of compromised IIoT equipments from the industrial control system to halt the spread of botnet and avert from further breaches. This research focuses on the development of botnet attack detection by combining deep learning methods with meta-heuristic optimization algorithms. The proposed method employs Grey Wolf Optimization (GWO) to optimally choose the attributes that can be utilized to classify the attacks. In classification step, Deep Variational Auto encoders (DVAE) are used to categorize the botnet attacks as benign or anomalous. In this work, N-BaIoT dataset with benign and anomalous attacks caused by Gafgyt and Mirai botnets are applied with the proposed methodology and the results are analyzed. The performance of the proposed DVAE-GWO is compared against the recent existing works to demonstrate the effectiveness of the proposed model. Performance metrics such as accuracy, precision, recall and F1 score are computed and it is observed that DVAE-GWO produces a higher accuracy of 98.8% than the other methods