Coordinated Scheduling and Spectrum Sharing via Matrix Fractional Programming

Kaiming Shen, Wei Yu, Licheng Zhao, Daniel P. Palomar

Abstract

Interference management is a fundamental issue in device-to-device (D2D) communications whenever the transmitter-and-receiver pairs are located in close proximity and frequencies are fully reused, so active links may severely interfere with each other. This paper devises an optimization strategy named FPLinQ to coordinate the link scheduling decisions among the interfering links, along with power control and beamforming. The key enabler is a novel optimization method called matrix fractional programming (FP) that generalizes previous scalar and vector forms of FP in allowing multiple data streams per link. From a theoretical perspective, this paper provides a deeper understanding of FP by showing a connection to the minorization-maximization (MM) algorithm. From an application perspective, this paper shows that as compared to the existing methods for coordinating scheduling in the D2D network, such as FlashLinQ, ITLinQ, and ITLinQ+, the proposed FPLinQ approach is more general in allowing multiple antennas at both the transmitters and the receivers, and further in allowing arbitrary and multiple possible associations between the devices via matching. Numerical results show that FPLinQ significantly outperforms the previous state-of-the-art in a typical D2D communication environment.

Introduction

In dense wireless networks where devices communicate directly (D2D), interference is a primary bottleneck. When multiple devices transmit simultaneously on the same frequency, the resulting noise floor can degrade signal quality for all users. Managing this requires solving a complex joint optimization problem involving:

  1. Scheduling: Determining which links should be active (a discrete combinatorial problem).
  2. Power Control: Optimizing the transmit power of active links.
  3. Beamforming: In Multiple-Input Multiple-Output (MIMO) systems, shaping signals to maximize data streams while suppressing interference.

Existing solutions like FlashLinQ or ITLinQ often rely on greedy heuristics and manually tuned parameters. Furthermore, standard optimization techniques like Block Coordinate Descent (BCD) frequently suffer from "premature turning-off," where a link is deactivated early and cannot be reactivated, even if it becomes globally beneficial later.

Methodology: Matrix Fractional Programming and MM

The core of the proposed framework is Matrix Fractional Programming (FP). This approach generalizes previous scalar and vector FP theories to the matrix domain, enabling the simultaneous optimization of multiple data streams per link.

A key theoretical contribution is the formal link between Matrix FP and the Minorization-Maximization (MM) algorithm. The authors demonstrate that the FP transforms (Matrix Quadratic and Matrix Lagrangian Dual) essentially construct surrogate functions that minorize the original non-convex objective.

The Minorization-Maximization (MM) principle: iteratively maximizing a surrogate function $g$ that lower-bounds the objective $f$.

By iterating between updating auxiliary variables and solving a weighted bipartite matching problem, FPLinQ avoids the "premature off" trap. Because it uses surrogate functions, it can reactivate links if the global interference environment improves, providing a more robust path to convergence.

Flexible Association and Network Modeling

Unlike traditional D2D models that assume a fixed one-to-one mapping between transmitters and receivers, FPLinQ supports a flexible association model. This allows transmitters to choose from multiple potential receivers, which is critical for dynamic mesh networks and complex interference environments.

Comparison between (a) Fixed Single Association and (b) Flexible Association models in D2D networks.

This flexibility, combined with the parameter-free nature of the algorithm, makes FPLinQ highly adaptable to various network topologies without requiring the "magic" tuning parameters needed by previous heuristics.

Numerical Results

The performance of FPLinQ was evaluated using numerical simulations in a 1km x 1km area with up to 900 D2D links, using the ITU-1411 channel model. Key findings include:

  • Sum-Rate Gains: FPLinQ achieved nearly double the sum-rate of FlashLinQ and significantly outperformed ITLinQ+.
  • Fairness: In MIMO scenarios, FPLinQ improved the 50th percentile link rate by over 50% compared to BCD methods when optimizing for log-utility.
  • Efficiency: The algorithm typically converges within 10 to 20 iterations, even in complex 8x8 MIMO configurations.

Conclusion

FPLinQ represents a significant advancement in interference management for dense MIMO D2D networks. By leveraging Matrix FP and its connection to MM, it provides a theoretically grounded, parameter-free method for joint scheduling and beamforming. Future work may address the challenges of decentralized implementation and the impact of imperfect Channel State Information (CSI) in real-world deployments.

Summary updated

Abstract

Interference management is a fundamental issue in device-to-device (D2D) communications whenever the transmitter-and-receiver pairs are located in close proximity and frequencies are fully reused, so active links may severely interfere with each other. This paper devises an optimization strategy named FPLinQ to coordinate the link scheduling decisions among the interfering links, along with power control and beamforming. The key enabler is a novel optimization method called matrix fractional programming (FP) that generalizes previous scalar and vector forms of FP in allowing multiple data streams per link. From a theoretical perspective, this paper provides a deeper understanding of FP by showing a connection to the minorization-maximization (MM) algorithm. From an application perspective, this paper shows that as compared to the existing methods for coordinating scheduling in the D2D network, such as FlashLinQ, ITLinQ, and ITLinQ+, the proposed FPLinQ approach is more general in allowing multiple antennas at both the transmitters and the receivers, and further in allowing arbitrary and multiple possible associations between the devices via matching. Numerical results show that FPLinQ significantly outperforms the previous state-of-the-art in a typical D2D communication environment.
BibTeX
@article{shen2019coordinated,
  author = {Kaiming Shen and Wei Yu and Licheng Zhao and Daniel P. Palomar},
  title = {{Coordinated Scheduling and Spectrum Sharing via Matrix Fractional Programming}},
  journal = {IEEE/ACM Trans. Networking},
  year = {2019},
  url = {https://openreview.net/forum?id=WpaTVrciFLn},
}
Figure