A cascaded channel-power allocation for D2D underlaid cellular networks using matching theory

Yiling Yuan, Tao Yang, Yuedong Xu, Hui Feng, Bo Hu

Abstract

As mobile data traffic continues to grow, Device-to-Device (D2D) communication has emerged as a critical technology for enhancing cellular network capacity. However, the interference caused by D2D pairs reusing cellular spectrum presents a significant challenge. This paper proposes a distributed, two-stage framework that decouples channel allocation and power control. By employing matching theory with externalities and a virtual pricing mechanism, the proposed scheme achieves near-optimal throughput with minimal signaling overhead.

Introduction

In D2D-underlaid cellular networks, nearby devices communicate directly while sharing frequencies with traditional cellular users (CUs). This spectrum reuse introduces complex interference management problems. Traditional centralized optimization is often computationally prohibitive (NP-hard) and requires global channel state information (CSI), which is impractical in large-scale deployments. This work addresses these challenges by decomposing the joint channel and power allocation problem into a cascaded distributed framework.

Methodology: The Cascaded Framework

The proposed solution operates in two distinct stages to manage complexity and information overhead.

The two-stage cascaded framework for channel allocation and power control.

Stage 1: Channel Allocation via Matching Theory

Channel selection is modeled as a many-to-one matching game between D2D pairs and cellular channels. Unlike standard matching models, this game incorporates externalities, where a D2D pair's utility depends on the interference from other pairs sharing the same channel. The algorithm uses a Gale-Shapley-based initialization followed by a swap-matching process to reach a strongly swap-stable state.

Stage 2: Power Control via Virtual Pricing

Following channel assignment, transmit power is optimized using a non-cooperative game. To protect primary CUs, the Base Station (BS) broadcasts a virtual price factor. D2D pairs adjust their power levels to maximize their individual rates while accounting for this "cost" of interference. A bisection search at the BS ensures the interference constraint remains tight and Pareto optimal.

Simulation and Performance Analysis

The framework was evaluated using Monte Carlo simulations in a cell with a 500m radius and 50m D2D links, employing standard pathloss and fast-fading models.

Comparison of the proposed scheme against brute-force and random allocation methods.

Key Results

  • Near-Optimality: The distributed algorithm achieves a sum rate within 10–20% of the computationally intensive brute-force method.
  • Efficiency: The scheme outperforms random allocation by over 35%, demonstrating the effectiveness of the matching-based approach.
  • Scalability: By allowing multiple D2D pairs per channel (many-to-one matching), the system achieves significantly higher throughput than orthogonal sharing schemes.
  • Cellular Protection: The pricing mechanism effectively maintains interference to cellular users below predefined thresholds.

Conclusion

This cascaded approach provides a practical, distributed solution for D2D interference management. By separating channel allocation from power control, it significantly reduces signaling overhead while maintaining high performance. Future work will focus on incorporating D2D-to-D2D interference awareness into the power control phase and evaluating performance in high-mobility environments.

Summary updated

BibTeX
@inproceedings{yuan2018cascaded,
  author = {Yiling Yuan and Tao Yang and Yuedong Xu and Hui Feng and Bo Hu},
  title = {{A cascaded channel-power allocation for D2D underlaid cellular networks using matching theory}},
  booktitle = {WCNC},
  year = {2018},
  doi = {10.1109/wcnc.2018.8377076},
}
Figure