A comprehensive survey of game theory models, applications, and future directions in cybersecurity
Year: 2026
Type: article
Abstract: This survey provides a comprehensive examination of game-theoretic solutions for cybersecurity, charting the field’s evolution from foundational models to the emerging paradigm of autonomous strategic defense. We systematize the literature by classifying core game-theoretic frameworks, including Stackelberg, Bayesian, and cooperative games, and examine their application to critical security tasks. For each model–application pairing, we grade the supporting evidence as theoretical, simulation-based, benchmark-based, or operationally deployed, showing that most cyber-domain approaches have not yet been validated in operational deployments. A central contribution of this work is a critical analysis of the theory–practice feedback loop, illustrated through in-depth case studies of seminal deployed physical-security systems such as ARMOR, PROTECT, and TRUSTS, whose lessons we relate to cyber defense. These examples reveal how real-world implementation challenges, including adversary bounded rationality and environmental uncertainty, have exposed the limitations of classical models and catalyzed fundamental advancements in the field. Building on this analysis, we synthesize the convergence of game theory and artificial intelligence, providing a structured examination of three critical frontiers: the strategic arms race in adversarial machine learning, the design of incentive-compatible security for federated learning, and the development of adaptive defense agents via multi-agent reinforcement learning in modern cyber gyms. Finally, we analyze the strategic challenges of collaborative security through the lens of dynamic coalition formation models. By identifying key research gaps and charting a forward-looking agenda, this survey aims to equip researchers and practitioners with a nuanced understanding of how to architect the next generation of intelligent, adaptive, and strategically robust cybersecurity systems.
Source: Discover Computing
Authors: Satish Korrapati, Ashutosh Bhatia
Institution: Birla Institute of Technology and Science, Pilani
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Keywords: Stackelberg game, incomplete information games, cooperative games, physical protection systems, armor, PROTECT, trust, environmental uncertainty, adversarial learning, federated learning security, multi-agent reinforcement learning, dynamic coalition formation, collaborative security, theory-practice gap
Open Access status: gold
Estimated APC paid: $1,520
OpenAlex ID: W7236265388
