Counterfactual Regret Minimization-Mixing for Noncooperative Stochastic Spectrum Games With Imperfect Information
IEEE Transactions on Networking · 2026 · 34 · 6230-6244
DOI: 10.1109/ton.2026.3693252 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Networking, 2026. Verfasst von Zuyuan Zhang und 3 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Gefördert von: ONR, USMA.
Abstract (Originalsprache)
Spectrum sharing is a key enabler for 5G/6G. We model decentralized spectrum access as a noncooperative, stochastic, imperfect-information extensive-form game and show that running standard Counterfactual Regret Minimization (CFR) independently at each user can exhibit persistent cycling rather than converging to a stable equilibrium. We provide a constructive multi-player example and a mapping analysis explaining why the induced regret-matching update need not be contractive in general multi-player, non-zero-sum settings. To address this, we propose CFR-M2, a lightweight regret-mixing scheme that periodically aggregates a small amount of cumulative regret (not policy parameters) across users at a configurable communication interval. Our analysis shows that inserting an infrequent regret-consensus step over any connected communication graph preserves no-(counterfactual)-regret while contracting inter-agent disagreement in regrets. Consequently, the Césaro-averaged play converges to an $\varepsilon $ –extensive-form coarse correlated equilibrium (EFCCE) with \[\varepsilon =\tilde O\!\left ({ \tfrac {1}{\sqrt {T}}}\right)+O\!\left ({ \tfrac {R_{\max }{\,}T_{\mathrm {comm}}}{(1-\rho){\,}T}}\right),\] where $\rho $ is the second-largest eigenvalue modulus of the mixing matrix. Under additional structure (e.g., a strongly monotone pseudo-gradient or a strongly convex potential), the EFCCE collapses to a unique Nash equilibrium. Experiments on synthetic networks and a 5G Dynamic Spectrum Sharing (DSS) scenario (WINNER II channel model and practical parameter settings) show that CFR-M2 improves convergence stability and system reward over CFR and several learning/game-theoretic baselines, while requiring only infrequent communications.
Förderung
- ONR
- USMA
Zitieren
APA
Zhang, Z., Liu, L., Bastian, N. D., & Lan, T. (2026). Counterfactual Regret Minimization-Mixing for Noncooperative Stochastic Spectrum Games With Imperfect Information. IEEE Transactions on Networking, 34, 6230-6244. https://doi.org/10.1109/ton.2026.3693252
Vancouver
Zhang Z, Liu L, Bastian ND, Lan T. Counterfactual Regret Minimization-Mixing for Noncooperative Stochastic Spectrum Games With Imperfect Information. IEEE Transactions on Networking. 2026;34:6230-6244. doi:10.1109/ton.2026.3693252
BibTeX
@article{Zhang2026,
author = {Zhang, Zuyuan and Liu, Lingjia and Bastian, Nathaniel D. and Lan, Tian},
title = {Counterfactual Regret Minimization-Mixing for Noncooperative Stochastic Spectrum Games With Imperfect Information},
journal = {IEEE Transactions on Networking},
year = {2026},
volume = {34},
pages = {6230--6244},
doi = {10.1109/ton.2026.3693252}
}
EUVIMED