Introduction
Intelligent Reflecting Surfaces (IRS) are a cornerstone of 6G research, praised for their ability to "program" the wireless environment by reflecting signals with controlled phase shifts. However, most existing literature assumes the IRS is an altruistic entity that always participates in communication at no cost. In real-world deployments, the Base Station (BS) and the IRS might belong to different operators. Activating thousands of IRS elements consumes power and accelerates hardware depreciation. This paper moves away from the idealistic assumption of "selfless" network components, modeling a market-based scenario where a selfish IRS operator charges a BS for its services.
System Model and Methodology
The authors propose an incentivized resource allocation framework using a Stackelberg game and a modular IRS architecture to jointly optimize resource allocation and pricing.
Modular IRS Architecture
Instead of controlling every single reflecting element individually, the authors propose a modular design. Elements are grouped into modules, each with its own controller. This allows the system to toggle entire modules on or off, significantly reducing the complexity of resource management and power consumption.
Stackelberg Game Framework
The interaction is modeled as a leader-follower game:
- The Leader (IRS): Sets a price for activating its reflecting modules to maximize its revenue.
- The Follower (BS): Given the price, decides which modules to activate and optimizes its transmit beamforming and the IRS's passive beamforming to maximize its net utility (data rate minus cost).
To solve the resulting non-convex mixed-integer optimization problem, the authors employ a Stackelberg game-based Alternating Direction Method of Multipliers (ADMM). They use a mixed $\ell_{1,2}$-norm to induce "group sparsity," which mathematically identifies which IRS modules should remain idle to save costs.
Key Insights and Results
Numerical simulations demonstrate that the proposed Stackelberg ADMM scheme consistently achieves higher utility for both the BS and the IRS compared to benchmarks like "random pricing" or "direct link only" (no IRS).
As shown in the figures, the algorithm successfully identifies the optimal number of active modules, proving that activating all elements is not always optimal when service fees are considered.
Price Sensitivity and Sum-Rate
An interesting dynamic emerges regarding price sensitivity: as the BS's own transmit power increases, its reliance on the IRS decreases. To remain competitive and ensure its modules are hired, the IRS operator must lower its prices as the BS becomes more "self-sufficient."
Even with the added cost of service, the IRS-aided system provides a significantly higher sum-rate for users compared to systems without an IRS, especially in scenarios with poor direct-link conditions.
Conclusion and Future Work
By treating the IRS as a service provider, this research bridges the gap between engineering optimization and socio-economic modeling, making it highly applicable to multi-operator 5G/6G ecosystems. Future work could extend this model to competitive games involving multiple IRS operators or multiple BSs, and account for practical constraints like imperfect Channel State Information (CSI) and hardware impairments.
Summary updated