Protecting cyber physical production systems using anomaly detection to enable self-adaptation

Abstract

The industrial world is going through its fourth revolution also known as Industry 4.0. Modern industrial processes leverage advanced IT technologies to increase productivity and often combine multiple system concepts such as Internet of Things (IoT), Cyber Physical Systems (CPS) and Cloud Computing. Cyber Physical Production Systems (CPPS) are key enablers of this revolution. In CPPS, raw materials, machines, and operations are interconnected to form a sophisticated network. Protecting them against advanced cyber-threats is a priority concern for the future implementation of Industry 4.0 applications. Any impairment of such systems can lead, in fact, to catastrophic damages resulting in a substantial financial loss for governments, companies, as well as endanger the safety of the society. The need for high availability and reliability of these systems is therefore the pillar guiding our research. This paper proposes the adoption of anomaly detection as a method to support self-adaptation in CPPS and to ensure flexibility, reliability, and protection of industrial environments against modern cyber threats. An anomaly detection mechanism can be employed to monitor, and learn the normal behavior of an industrial system, and to generate alerts when the observed events indicate abnormal activities. On this concept we base our work, and we demonstrate how timely identifying critical security events can enable, through the self-adaptation (e.g., triggering automatic configuration changes), an efficient protection of the CPPS against advanced threats, and an effective containment of their effects.