Reveling misleading information for defenders and attackers in repeated Stackelberg Security Games

Critical infrastructure protection has become a major problem in today's economy. Furthermore, the ongoing attacks on the population have piqued interest in resource allocation models in the face of societal dangers. Attacker–Defender Stackelberg Security Games (SSGs) have emerged as a critical field of study and development for resolving this issue. A game-theoretic model for SSGs with incomplete information is presented in this study. The objective is to reduce the knowledge and ingenuity of attackers when it comes to selecting a target, location, and time for an assault. For security resource allocation, we present a Bayesian-Stackelberg game-theoretic framework. The defenders’ diverse preferences create a decision-making dilemma, which needs the development of a system to devise strategies that encourages coordination. We construct an incentive-compatible optimum mechanism that maximizes benefits while incentivizing participants to implement the offered strategies. Our findings demonstrate that is possible to compute a mechanism able to obtain effective defensive coordination in security games under certain restrictions. In this approach, attackers and defenders learn their behavior by seeing private information in a Markov process-restricted game. The artificial intelligence technology chosen to perform the learning process is Reinforcement Learning (RL) approach. We provide an algorithm and evaluate the repeating game using myopic players as attackers and defenders. Experiments on SSGs can be managed using a random walk technique. The suggested framework's efficacy and efficiency are demonstrated using a numerical example.