Modeling and Analysis of MDP-based Security Risk Assessment System for Smart Grids

Abstract

The recent advances in communications and networking technologies have made the electricity service provisioning more convenient, affordable and manageable through the realization of Smart Grids. However, these cyber-physical systems have become vulnerable to security attacks leading to service disruptions. To address this challenge, it is crucial to design, model and implement a robust security risk management solution to protect the Smart Grids in the event of a security attack. In this paper we propose, implement and evaluate a Probabilistic Model Checking (PMC) system to model the security attack process. In particular, we model the attacker behavior through a Markov Decision Process (MDP) which non-deterministically and probabilistically captures the attack process in a Smart Grid system consisting of eight important components. Two major scenarios are presented, considering the non-persistence and persistence attacker behaviors. The parameters used in the PRISM tool to evaluate the robustness of the proposed model were the Probability of successful attack, Attack costs, and Number of attack attempts. The proposed implementation provides useful insights into the best possible defense or preventive mechanisms that can be adopted by the security manager in the event of a specific security attack.