Abstract:Deep learning has recently emerged as a disruptive technology to solve challenging radio resource management problems in wireless networks. However, the neural network architectures adopted by existing works suffer from poor scalability, generalization, and lack of interpretability. A long-standing approach to improve scalability and generalization is to incorporate the structures of the target task into the neural network architecture. In this paper, we propose to apply graph neural networks (GNNs) to solve large-scale radio resource management problems, supported by effective neural network architecture design and theoretical analysis. Specifically, we first demonstrate that radio resource management problems can be formulated as graph optimization problems that enjoy a universal permutation equivariance property. We then identify a class of neural networks, named \emph{message passing graph neural networks} (MPGNNs). It is demonstrated that they not only satisfy the permutation equivariance property, but also can generalize to large-scale problems while enjoying a high computational efficiency. For interpretablity and theoretical guarantees, we prove the equivalence between MPGNNs and a class of distributed optimization algorithms, which is then used to analyze the performance and generalization of MPGNN-based methods. Extensive simulations, with power control and beamforming as two examples, will demonstrate that the proposed method, trained in an unsupervised manner with unlabeled samples, matches or even outperforms classic optimization-based algorithms without domain-specific knowledge. Remarkably, the proposed method is highly scalable and can solve the beamforming problem in an interference channel with $1000$ transceiver pairs within $6$ milliseconds on a single GPU.

Graph Neural Networks for Quality of Service Improvement in Interference-Limited Ultra-Reliable and Low-Latency Communications

Interference-Limited Ultra-Reliable and Low-Latency Communications: Graph Neural Networks or Stochastic Geometry?

Recurrent Neural Network Based Collaborative Filtering for QoS Prediction in IoV

Design of Retransmission Mechanism for Decentralized Inference with Graph Neural Networks

Joint URLLC Traffic Scheduling and Resource Allocation for Semantic Communication Systems

Deep Reinforcement Learning Based Massive Access Management for Ultra-Reliable Low-Latency Communications

Multicell Power Control under QoS Requirements with CNet

Privacy-Preserving Decentralized Inference with Graph Neural Networks in Wireless Networks

Optimal Wireless Resource Allocation with Random Edge Graph Neural Networks

Learning to Optimize Resource in Dynamic Wireless Environment Via Meta-Gating Graph Neural Network

Machine-Learning-Based Parallel Genetic Algorithms for Multi-Objective Optimization in Ultra-Reliable Low-Latency WSNs

Deploying Graph Neural Networks in Wireless Networks: A Link Stability Viewpoint

Reliability-Optimized User Admission Control for URLLC Traffic: A Neural Contextual Bandit Approach

Distributed Learning Framework for eMBB-URLLC Multiplexing in Open Radio Access Networks

Graph Neural Networks for Scalable Radio Resource Management: Architecture Design and Theoretical Analysis

Deep Learning for Ultra-Reliable and Low-Latency Communications in 6G Networks

Secure and Reliable Downlink Transmission for Energy-Efficient User-Centric Ultra-Dense Networks: An Accelerated DRL Approach

Ultra-Reliable and Low-Latency Wireless Communications in the High SNR Regime: A Cross-Layer Tradeoff

Graph Convolutional Network Enabled Power-Constrained HARQ Strategy for URLLC

Knowledge-Driven Resource Allocation for Wireless Networks: A WMMSE Unrolled Graph Neural Network Approach

Meta-Learning Empowered Graph Neural Networks for Radio Resource Management