A Machining Process Invariant Approach to Cyber-Attack Detection on Cyber-Physical Manufacturing Systems

Abstract

With more and more Internet of Things (IoT) devices being used in manufacturing, cyber-physical manufacturing systems (CPMSs) have become simpler targets for hackers. There have been major developments in the area of CPMS intrusion detection recently. Nevertheless, the present approaches entail substantial expenses when gathering past data to train detection models, which are customized for certain machining situations. These procedures are put to the test by the ever-changing nature of machining in practice. In this research, we lay the framework for detection by detailing our finding that the CPMS machining code contains a whole machining process. Using the Machining Process Invariant of the machining code, we provide MPI-CNC, an intrusion detection technique. Automating the analysis of machining codes to build machining process rules and important parameter sets is possible with MPI-CNC. As rules for detection, they are very useful. After then, MPI-CNC seeks out the CPMS's runtime state and checks it against the detection criteria to determine how the cyberattack behaved. We ran 10 actual machining operations using MPI-CNC on two FANUC CNC machines to see how well it performed. The results of the experiments proved that MPI-CNC is quite versatile. While keeping regular machining processes going, MPI-CNC demonstrated it could identify cyber-attacks on CPMS with more accuracy than state-of-the-art detection systems.