Generative Diffusion Models for Resource Allocation in Wireless Networks

2025-04-28

AI Summary

This work addresses the joint optimization of long-term utility maximization and strict Quality-of-Service (QoS) constraints in wireless networks. Methodologically, we propose a stochastic policy learning framework based on Generative Diffusion Models (GDMs)—the first application of GDMs to wireless resource allocation—where a supervised policy-learning paradigm imitates expert solutions; Graph Neural Networks (GNNs) encode network topology for cross-topology generalization; and sequential sampling approximates the optimal policy. Evaluated on multi-user interference channel power control, our approach significantly outperforms conventional reinforcement learning and heuristic methods. It achieves near-optimal long-term utility while rigorously satisfying hard QoS constraints. The framework combines theoretical soundness—grounded in diffusion-based policy optimization and graph representation learning—with practical deployability, enabling scalable, topology-agnostic resource management in dynamic wireless environments.

Technology Category

• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Planning with Markov Models (MDPs, POMDPs)
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Content-based information diffusion
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

This paper proposes a supervised training algorithm for learning stochastic resource allocation policies with generative diffusion models (GDMs). We formulate the allocation problem as the maximization of an ergodic utility function subject to ergodic Quality of Service (QoS) constraints. Given samples from a stochastic expert policy that yields a near-optimal solution to the problem, we train a GDM policy to imitate the expert and generate new samples from the optimal distribution. We achieve near-optimal performance through sequential execution of the generated samples. To enable generalization to a family of network configurations, we parameterize the backward diffusion process with a graph neural network (GNN) architecture. We present numerical results in a case study of power control in multi-user interference networks.

Problem

• Learning stochastic resource allocation policies with diffusion models
• Maximizing ergodic utility under QoS constraints in wireless networks
• Generalizing optimal allocation via GNN-parameterized diffusion for varying networks

Innovation

• Supervised training algorithm with generative diffusion models
• Parameterized backward diffusion using graph neural networks
• Sequential execution for near-optimal resource allocation