Using machine learning algorithm for detection of cyber-attacks in cyber physical systems

Abstract

Network integration is common in cyber-physical systems (CPS) to allow for remote access, surveillance, and analysis. They have been exposed to cyberattacks because of their integration with an insecure network. In the event of a violation in internet security, an attacker was able to interfere with the system's functions, which might result in catastrophic consequences. As a result, detecting breaches into mission-critical CPS is a top priority. Detecting assaults on CPSs, which are increasingly being targeted by cyber criminals and cyber threats, is becoming increasingly difficult. Machine Learning (ML) and Artificial Intelligence (AI) have the potential to make these the worst of moments, but it may also be the finest of times. There are a variety of ways in which AI technology can aid in the growth and profitability of a variety of industries. Such data can be parsed using ML and AI approaches in designed to check attacks on CPSs. Hence, in this paper, we propose a novel cyberattack detection framework by integrating AI and ML (ML) methods. Here, initially we collect the dataset from the CPS database and preprocess the data using normalization for removal of errors and redundant data. The features are extracted using Linear Discriminant Analysis (LDA). We have proposed Self-tuned Fuzzy Logic-based Hidden Markov Model (SFL-HMM) with Heuristic Multi-Swarm Optimization (HMS-ACO) algorithm for detection of the cyberattacks. The proposed method is evaluated using the MATLAB simulation tool and the metrics are compared with existing approaches. The results of the experiments reveal that the framework is more successful than traditional strategies in achieving high degrees of privacy. Furthermore, in terms of detection rate, false positive rate, and computing time, the framework beats traditional detection algorithms.