Introduction
As cellular networks expand, their energy consumption and resulting carbon footprint have become major operational and environmental challenges. Powering base stations (BSs) with renewable energy sources (RPS) like solar or wind is a promising solution, but it introduces two significant hurdles: intermittency and conflicting interests. In a decentralized market, the power supplier (RPS) and the network operator (BS) are often independent entities, each seeking to minimize their own costs. This paper proposes a game-theoretic framework to manage renewable energy allocation, modeling the system as an "M/M/1 make-to-stock" queue to coordinate decentralized players for near-optimal performance.
System Model and Methods
The authors employ a mix of queuing theory, game theory, and mechanism design to bridge the gap between independent actors. The core of the approach is the M/M/1 Make-to-Stock Queue model. Unlike standard queues where customers wait for service, this model maintains an energy inventory. The RPS acts as the server replenishing the stock at a rate $\mu$, while incoming mobile connection demands act as "customers" consuming it.
The interaction is modeled as a noncooperative game where the BS chooses its energy inventory level ($s$) and the RPS chooses its supply rate ($\mu$). To address the inherent inefficiency of the Nash Equilibrium (NE), the authors introduce linear contracts (transfer payments) to align incentives toward a socially optimal solution. Furthermore, for scenarios with a monopolistic RPS and multiple BSs, an Incentive Compatible (IC) mechanism is proposed to ensure BSs report their true energy needs.
Key Insights and Results
A primary insight is the "Make-to-Stock" analogy, which allows for balancing the cost of holding energy (battery self-discharge/capital) against the cost of "backlogging" (dropped or delayed connections). The study found that standard proportional allocation is not Pareto optimal in this context because it ignores varying arrival rates and BS sensitivities.
Key results include:
- Uniqueness of Equilibrium: The decentralized game has a unique Nash Equilibrium, providing a stable baseline.
- Coordination Success: Linear contracts successfully reduce total costs compared to the standard NE, reaching the centralized optimal state.
- Cost Savings: Simulations using industry-standard parameters (e.g., 24 W per microcell connection) show significant savings under light connection demand. The proposed IC mechanism performs close to the social optimum as renewable capacity increases.
- Hybridization: The model demonstrates that BSs can effectively toggle between the grid and renewables to "shave" peak energy bills even under heavy traffic.
Limitations and Future Work
The current model assumes a "one-shot" game with fixed pricing and stationary users. Future extensions could explore repeated game models to capture dynamic market behavior, incorporate dynamic pricing based on real-time supply/demand, and account for user mobility and handovers between base stations.
Source: Li, D., et al. (2015). Decentralized Energy Allocation for Wireless Networks with Renewable Energy Powered Base Stations. arXiv:1506.05549v1.
Summary updated