Abstract
As mobile data demands surge, cellular operators are increasingly utilizing the unlicensed 5GHz spectrum via LTE-Unlicensed (LTE-U). However, the coexistence of LTE's scheduled protocol with Wi-Fi's contention-based protocol often leads to significant interference. This paper proposes a dual-game theoretic framework to manage this "noisy neighbor" problem. By combining a cooperative Nash Bargaining Game for time-sharing with a one-sided matching game for resource allocation, the framework ensures a fair and efficient distribution of spectrum. Results indicate that the proposed method boosts cellular capacity while protecting Wi-Fi performance more effectively than standard Listen-Before-Talk (LBT) protocols.
Introduction: The Coexistence Challenge
The explosion of mobile data traffic has pushed cellular network operators (CNOs) toward the unlicensed 5GHz spectrum, traditionally dominated by Wi-Fi. This integration, known as LTE-U, faces a fundamental protocol clash: LTE is designed for exclusive, scheduled access, whereas Wi-Fi relies on "polite" contention-based access. Without coordination, aggressive LTE-U deployments can "drown out" Wi-Fi, causing severe performance degradation.
The problem is further complicated in multi-operator scenarios where multiple CNOs compete for the same unlicensed space, creating a three-way interference dynamic between different LTE operators and existing Wi-Fi networks.
Methodology: A Dual-Game Framework
To address this, the authors formulate a Mixed Integer Non-Linear Programming (MINLP) problem to maximize the LTE-U sum-rate while respecting Quality-of-Service (QoS) and Wi-Fi protection constraints. Given the NP-hard nature of MINLP, the problem is decomposed into two iterative sub-problems:
- Time-Sharing via Nash Bargaining Game (NBG): A cooperative game determines the fraction of time the unlicensed channel is shared between LTE and Wi-Fi. The NBG ensures a "win-win" by guaranteeing Wi-Fi a minimum throughput (disagreement point) and maximizing mutual utility for the remaining time.
- Resource Allocation via One-Sided Matching: Once the time-slice is set, a "House Allocation Model" from matching theory assigns unlicensed subchannels to users. Subchannels act as agents with preferences based on potential data rates, leading to a Pareto-optimal allocation.
A centralized arbitrator coordinates these games until the system converges to a stable local optimum.
Key Insights
- Cooperation over Sensing: Explicit game-theoretic cooperation provides more stable fairness in dense environments than simple carrier sensing (LBT).
- Wi-Fi as a Partner: By treating Wi-Fi as a bargaining partner with guaranteed performance floors, the system achieves better protection for Wi-Fi users than non-cooperative "polite" protocols.
- Efficient Decomposition: Splitting the problem into time-sharing and matching makes the complex optimization computationally feasible for real-world network controllers.
Results and Performance
Simulation results in a 150-meter radius environment with varied QoS profiles (HD video, VoIP, etc.) highlight the framework's efficacy:
- Throughput Gains: The proposed method achieved per-user rates approximately 2.7% to 24.5% higher than baseline LTE-U and LTE-A methods.
- QoS Satisfaction: In single-channel scenarios, the percentage of "unsatisfied" users (those not meeting QoS) dropped significantly from 60% (LTE-A) to 39%.
- Fairness and Protection: The approach maintained high Jain’s Fairness Index scores and shielded Wi-Fi users more effectively than conventional LBT schemes, providing higher normalized throughput for WAPs.
Conclusion and Future Directions
While the proposed framework offers significant gains, it currently relies on a centralized arbitrator and requires information exchange between operators, which may pose implementation and privacy challenges. Future research will explore decentralized mechanisms and the integration of machine learning to dynamically predict traffic and interference patterns.
Summary updated