Inter-Operator Spectrum Sharing from a Game Theoretical Perspective

Mehdi Bennis, Samson Lasaulce, Mérouane DebbahEURASIP J. Adv. Signal Process 2009 · DOI

Abstract

Spectrum scarcity and underutilization pose significant challenges for modern wireless networks. This paper explores inter-operator spectrum sharing through game theory, comparing simultaneous non-cooperative games (Nash) with hierarchical games (Stackelberg). The authors demonstrate that a leader-follower structure resolves the unpredictability of multiple Nash equilibria and consistently improves achievable rates for all participating operators.

Introduction: The Challenge of Shared Spectrum

Traditionally, frequency bands are licensed exclusively to single operators, leading to inefficient "underutilization" where some bands remain idle while others are congested. Spectrum sharing allows multiple operators to coexist in the same band, but it introduces the problem of mutual interference. When operators act purely selfishly and simultaneously, they often reach a Nash Equilibrium that is Pareto-inefficient and unpredictable due to the existence of multiple possible outcomes.

Methodology: Nash vs. Stackelberg

The authors analyze two primary frameworks to address this interference:

  1. Non-Cooperative Nash Game: Operators use "Iterative Water-Filling," adjusting their own power based on observed interference from others. The paper rigorously maps the "Nash regions," identifying where the system's behavior becomes unstable or unpredictable.
  2. Hierarchical Stackelberg Game: A hierarchy is introduced with a "Leader" (Primary operator) and a "Follower" (Secondary operator). The Leader anticipates the Follower's reaction to its power allocation, leading to a unique and more efficient equilibrium.
Nash equilibrium regions for a two-operator, two-carrier system, illustrating the complexity of the strategy space.

To scale this to multi-operator scenarios, the authors propose a distributed algorithm using Lagrangian duality and coordinate descent, comparing it against a centralized sum-rate maximization benchmark solved via Geometric Programming.

Key Results and Performance

Numerical simulations using synthetic Rayleigh fading channels demonstrate that the Stackelberg approach provides a significant performance boost over standard non-cooperative competition.

Average achievable rate for both users vs. SNR, comparing the Stackelberg game against Best/Worst Nash equilibria and the centralized sum-rate.

As shown in the simulation results, the Stackelberg approach consistently outperforms the "Worst Nash" equilibrium and stays above the "Best Nash" in most SNR regimes. Furthermore, the proposed distributed algorithm converges reliably and performs remarkably close to the theoretical centralized optimum, which requires perfect global knowledge.

Insights and Future Directions

The transition from "chaos" (simultaneous competition) to "hierarchy" (Stackelberg) acts as a stabilizer. By moving to a Stackelberg model, the equilibrium becomes unique, removing the unpredictability problem. Interestingly, even the "follower" often benefits from the leader's strategic optimization compared to a purely selfish Nash scenario.

However, practical implementation faces hurdles, such as the requirement for full Channel State Information (CSI) across operators. Future work may extend these hierarchical models to femtocell networks, where small base stations must share spectrum with macrocell leaders.

Reference: Bennis, M., Lasaulce, S., & Debbah, M. (2009). Inter-Operator Spectrum Sharing from a Game Theoretical Perspective. EURASIP Journal on Advances in Signal Processing.

Summary updated

BibTeX
@article{bennis2009inter,
  author = {Mehdi Bennis and Samson Lasaulce and Mérouane Debbah},
  title = {{Inter-Operator Spectrum Sharing from a Game Theoretical Perspective}},
  journal = {EURASIP J. Adv. Signal Process.},
  year = {2009},
  doi = {10.1155/2009/295739},
}
Figure