Uncovering cyber-threats to nuclear system sensing and observability

Cyber-physical systems are engineered systems that integrate physical processes and computational resources. But, by integrating cyber and physical worlds, the physical assets are vulnerable to cyber-attack. Two things are of importance for the security of cyber-physical assets: access to control inputs by the attacker, and the ability of an attacker to mask inputs. This combination of attacker control and masking measurements can allow an attacker to cause significant damage to a system while remaining undetected. By masking certain measurement signals, an attacker may affect the observability of the system and create a condition where part of the state space is unobservable, meaning that it is impossible to reconstruct those states. This is called an observability attack. This paper presents a technique for analyzing observability attacks. How an attacker can design an attack to maximize the impact on the unobservable states while minimizing the possibility of detection is discussed. Criteria for maintaining a stealthy attack are given, and a design method is provided. For a nuclear balance of plant system, combinations of sensor omissions are analyzed to find an observability attack with maximum impact and minimum detection. An appropriate attack input signal is created, an attack is simulated, and the system response is shown.