Markov Decision Process based Cost-Benefit Analysis of Cybercrime and Cyberdefense systems
D. Ezhilarasan, N.P.G. Bhavani, Bhanu Sekhar Guttikonda, Haeedir Mohameed, Hirald Dwaraka Praveena
Abstract
Over the past few years, Artificial Intelligence (AI)-driven decision-making systems optimize processes and improve efficiency but requires cost-benefit analysis to balance performance and expenses. However, traditional approaches for cost-benefit analysis had faced challenges like lack of sequential optimization and static decision making. Therefore, this research proposes Markov Decision Process (MDP) for cost benefit analysis in cybercrime and cyberdefense. Initially, intelligent information gathering is performed for collecting cybercrime-related data like incidents, vulnerability repositories, and financial impacts. Then, parameters for the unlicenced goods pricing model are determined for estimating the pricing and rewards which form MDP basis. After that, costs and benefits for defenders are analyzed by considering GDPR fines, legal fees, and defence rewards, weighing potential financial gains against legal risks are assessed for attackers. Finally, the MDP-based optimization determines the optimal defence strategies by maximizing long-term rewards and minimizing cyberattack impacts. The proposed MDP achieved better results in terms of Incremental Cost-Effectiveness Ratio (ICER) as -25,325 per QALY gain and probability of the intervention being cost saving was estimated at 90% when compared with existing Stepped-Wedge Cluster Randomized Controlled Trial (SWCRCT).