GAME-THEORETIC OPTIMIZATION MODEL FOR SELECTING SECURITY CONTROLS IN DISTRIBUTED INFORMATION SYSTEMS

Abstract

The article proposes a game-theoretic optimization model for selecting security controls for distributed information systems (DIS) under conditions of targeted cyber opposition. The relevance of this study is driven by the increasing architectural complexity of distributed information systems, limited security resources on the defense side, and the presence of a rational or boundedly rational adversary acting strategically. In contrast to existing optimization methods and traditional game-theoretic models, the proposed model integrates a network-based description of the DIS, the formalization of security control selection, and a Bayesian interpretation of the attacker's behavior. In this study, the DIS is represented as a directed graph with heterogeneous node criticality, accounting for the network effects of attack propagation. Security controls are modeled as discrete alternatives subject to the defender's limited budget. To assess the consequences of the conflict, a total Bayesian expected loss function is introduced. This function aggregates local damages, taking into account the prior probabilities of attacker types and their attack scenarios against the DIS. The optimal selection of a defense strategy is formulated as a problem of minimizing total Bayesian expected losses, which possesses a game-theoretic interpretation. To validate the model's functionality, computational experiments were conducted using the Python programming language within the PyCharm IDE. The obtained results confirmed the non-linear nature of the dependency between loss levels and the security budget, demonstrating the advantages of strategically balanced solutions over locally optimal approaches. These findings confirm the expediency of applying the game-theoretic optimization method for decision support in the field of distributed information systems cybersecurity.