Integrated Clinical Environment Security Analysis Using Reinforcement Learning

Many communication standards have been proposed recently and more are being developed as a vision for dynamically composable and interoperable medical equipment. However, few have security systems that are sufficiently extensive or flexible to meet current and future safety requirements. This paper aims to analyze the cybersecurity of the Integrated Clinical Environment (ICE) through the investigation of its attack graph and the application of artificial intelligence techniques that can efficiently demonstrate the subsystems' vulnerabilities. Attack graphs are widely used for assessing network security. On the other hand, they are typically too huge and sophisticated for security administrators to comprehend and evaluate. Therefore, this paper presents a Q-learning-based attack graph analysis approach in which an attack graph that is generated for the Integrated Clinical Environment system resembles the environment, and the agent is assumed to be the attacker. Q-learning can aid in determining the best route that the attacker can take in order to damage the system as much as possible with the least number of actions. Numeric values will be assigned to the attack graph to better determine the most vulnerable part of the system and suggest this analysis to be further utilized for bigger graphs.