Zero-Day Attacks Detection Using an Analysis of Mobile Robot Motor Primitives

In recent years, significant increase in threats and vulnerabilities is observed for different cyber-physical systems. Often attacks on cyber-physical systems aim to covertly destroy systems. The greatest threat is posed by zero-day attacks that mask their actions as internal malicious activities. Such masking is particularly effective for mobile robotic systems. Efficiency of intrusion detection systems for mobile robotic systems depends critically on efficiency of separation of internal malicious activities and zero-day attacks. We consider an approach that allows us to separate internal activity and external intrusions for mobile robotic systems. In particular, we consider an approach that is based on active exploration of internal states by mobile robotic systems. For effective active exploration of internal states, we propose an approach to the elimination of collaborative activity. We consider an approach to detecting malicious activity that is based on an analysis of discrete and rhythmic mobile robot motor primitives.