Securing Industrial Control Systems (ICS) Through Attack Modelling and Rule-Based Learning

In recent years, Industrial Control Systems (ICS) have been targeted through a range of cyber-attacks that attempt to infiltrate and disrupt process controls of industrial control systems. This is accomplished through direct manipulation of sensor and actuator states with consequent impact on critical infrastructure such as water treatment and power generation systems. Such cyber-physical attacks can be prevented through rule-based learning techniques. Our work comprises the design of a generator framework that employs the association rule mining technique to automatically generate attack variants and system invariants for an ICS. These generated attack variants and system invariants are validated through experimentation with results showing promise.