The behavioral approach to diagnostics of cyber-physical systems

Computer controlled machinery is perceived as a cyber-physical system. Such systems are vulnerable to cyber attacks of ever-growing sophistication with potentially severe consequences. To address this threat advanced security/status monitoring measures are to be developed and deployed within the framework of industrial control systems. A system diagnostic approach based on modeling and assessment of the overall behavior of the cyber-physical system utilizing system call data of its dedicated computer is presented. The behavior based anomaly/misuse detection procedures, described herein, have resulted in a system diagnostic technology capable of timely detection, identification and prediction of hardware malfunction caused by cyber attacks on its computer control systems. The approach is implemented with a numerically efficient system call processing algorithms for off-line functionality extraction and on-line functionality matching. Implemented in software, our approach was deployed and tested for a flight control computer operating an advanced flight simulator.