Risk-Aware Satisfaction Equilibria in Resource Sharing Games

• 成果类型: Article
• 署名作者: Promponas, Panagiotis; Pelekis, Christos; Tsiropoulou, Eirini Eleni; Papavassiliou, Symeon
• 署名单位: National Technical University of Athens; University of New Mexico
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2024.3426388
• 发表日期: 2025
• 页码: 278-292
• 关键词: games; resource management; dynamic scheduling; quality of service; Nash equilibrium; INVESTMENT; uncertainty; Best response dynamics (BRD); common pool resources (CPR); game theory; Resource sharing; satisfaction equilibria; supermodular games

摘要

The fragile common pool of resource (CPR) game is an instance of a resource sharing game where a common resource, which is prone to failure due to overuse, is shared among several players. Players have a fixed initial endowment and are faced with the task of investing in the common resource without forcing it to failure. The return from the common resource is subject to uncertainty and stochasticity, and is perceived by the players in a prospect-theoretic manner based on their behavioral characteristics. The Fragile CPR Game admits a unique Nash equilibrium (NE), and it has been shown in the literature that the best response dynamics converge to the NE. In this article, we look at the Fragile CPR Game through the lenses of Games in Satisfaction Form. We refer to the corresponding game as the Fragile SAT-CPR Game. Our main and novel result states that the Fragile SAT-CPR Game admits the optimal satisfaction equilibrium (OSE) and that the proposed risk minimizing dynamics, which are based on the best response dynamics, converge to the OSE. This equilibrium point results in minimizing the probability that the CPR collapses, while the players can obtain the same payoffs as in the NE point. Numerical evaluations indicate that in the OSE the probability that the common resource collapses can be decreased by approximately 95% compared to corresponding one in the NE case, while at the same time the players enjoy the same utility values. Our proofs employ and exploit concepts from the theory of Constrained S-modular Games.