Optimization of Cyber Defense Strategies in Smart Grids via Adaptive Autonomy and Knapsack Formulation

The increasing digitalization of smart mobility and transportation ecosystems has intensified their dependence on resilient and cyber-secure energy infrastructures. As intelligent transport systems, EV charging networks, and digitally managed logistics platforms rely fundamentally on stable smart grid operations, cyber disruptions to power distribution networks can directly compromise mobility continuity and operational safety. This research develops a two-player, zero-sum Stackelberg game model that embeds Adaptive Autonomy (AA) into the cybersecurity decision-making process of smart grids supporting mobility-critical infrastructure. The proposed framework models the interaction between an attacker, aiming to maximize customer interruption costs (CIC) under budget constraints, and a defender, seeking to minimize both CIC and autonomy-implementation costs. The defender’s strategy dynamically regulates the Level of Automation (LOA), optimizing resilience through adaptive trade-offs between human supervision and autonomous control. Both decision spaces are formulated as knapsack problems and solved using a Particle Swarm Optimization (PSO) algorithm, applied to the IEEE Billinton Four-Bus Test System. Simulation results demonstrate that adaptive autonomy enables flexible resource allocation, which significantly reduces interruption costs. Moreover, a cost-sensitivity analysis reveals that when autonomy implementation exceeds approximately 25% of average payout costs, its marginal benefit diminishes sharply. The proposed model provides a rigorous quantitative foundation for balancing economic efficiency and cyber-resilience in smart distribution systems, offering actionable insights for utilities and policymakers. By linking game-theoretic defence modelling with human–automation dynamics, this work introduces a novel paradigm for adaptive, cost-aware cybersecurity management in cyber-physical energy infrastructures.