Reinforcement Learning-Based Attack Graph Analysis for Wastewater Treatment Plant

Automation frequently needs less human engagement, encouraging reliance on continually operating machinery and automated processes that perform a range of functions. The resultant predictable, repeated behavior can be exploited. Since the Internet of Things (IoT) has been included in aforesaid automatic operations, these Cyber-Physical Systems (CPSs) are exposed to cyberattacks, making it difficult to stop them and identify their patterns. Even though running Wastewater Treatment Plant (WTPs) might be difficult, they are necessary since both drinking water and water that may be recycled are in low supply. Thus, increasing their vulnerabilities to assaults caused by the exploitation of flaws while doing so. To secure such CPSs, one must be aware of system weaknesses and potential exploitation. This article examines the attack graph of the treatment method, and its flaws to avoid similar incidents and lessen the harm. A Q-learning-based attack graph inquiry method is also presented, where the agent is considered to be the adversary and the attack graph that was formed mimics Q-environment. This method can help in figuring out the optimal path an attacker can follow to do the most damage with the fewest amount of actions.