Graph Signal Diffusion Models for Wireless Resource Allocation

Yigit Berkay Uslu; Samar Hadou; Shirin Saeedi Bidokhti; Alejandro Ribeiro

eess.SPApr 6, 2026

Graph Signal Diffusion Models for Wireless Resource Allocation

Authors: Yigit Berkay Uslu, Samar Hadou, Shirin Saeedi Bidokhti, Alejandro Ribeiro

Organizations: Dept. of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA, USA

Abstract

We consider constrained ergodic resource optimization in wireless networks with graph-structured interference. We train a diffusion model policy to match expert conditional distributions over resource allocations. By leveraging a primal-dual (expert) algorithm, we generate primal iterates that serve as draws from the corresponding expert conditionals for each training network instance. We view the allocations as stochastic graph signals supported on known channel state graphs. We implement the diffusion model architecture as a U-Net hierarchy of graph neural network (GNN) blocks, conditioned on the channel states and additional node states. At inference, the learned generative model amortizes the iterative expert policy by directly sampling allocation vectors from the near-optimal conditional distributions. In a power-control case study, we show that time-sharing the generated power allocations achieves near-optimal ergodic sum-rate utility and near-feasible ergodic minimum-rates, with strong generalization and transferability across network states.

Figures & tables

Figure 1: Primal-dual expert policies. Top-left panel shows the evolution of the dual multipliers associated with the ergodic minimum-rate constraints of selected receivers for an example network

{\mathbf{H}}

in the power-control problem of Section 3 . The golden shaded region marks the convergence regime. Bottom-left panel plots the per-iteration fraction of violated constraints against the worst-case constraint slack, averaged over a

2000

-iteration moving window. Although the iterates

{\mathbf{x}}^{\dagger}(\boldsymbol{\lambda}_{k})

remain infeasible for at least 20% of the receivers, the randomized policy

\widehat{{\mathcal{D}}}_{{\mathbf{x}}}(\cdot\,|\,{\mathbf{H}})

is near-optimal and feasible in the average-sense [cf. ( 11 ), ( 12 )]. The long transient preceding convergence is evident from both panels. The panels on the right show several 2D slices of the allocation vectors drawn from the expert policy in the convergence regime, with samples from the transient regime shown as faint markers. The top-row slices correspond to the receivers plotted in the top-left panel. The orange overlays show the corresponding slices for samples generated by the learned policy

\mathcal{D}_{{\mathbf{x}}}(\cdot\,|\,{\mathbf{H}};\boldsymbol{\theta}^{*})

, which closely matches the experts.

Figure 3: Generalization of U-GNN policies across

f_{\min}

(QoS) levels . The model is trained on

f_{\min}\in\{0.4,0.5,0.6,0.7,0.8\}

and evaluated at intermediate

f_{\min}

levels.

Figure 4: Size-transferability of U-GNN policies. The model is trained on networks of size

N=400

and four density levels

\nu\in\{6.6,8.2,10.1,11.9\}

tx-rx pairs/

\text{\,}{\mathrm{km}}^{2}

, and evaluated on networks of varying sizes at similar densities.

Wireless Resource Allocation

Wireless resource allocation assigns spectrum, power, beams, and computing capacity to meet throughput, latency, fairness, and energy goals.

See all 174 papers

Generative Diffusion Models of Stochastic Graph Signals
Jul 7, 2026Yiğit Berkay Uslu, Samar Hadou, Sergio Rozada +2Graph Neural Networks Financial Forecasting
Dept. of Electrical and Systems Eng., University of Pennsylvania, Philadelphia, PA 19104 USA · Department of Signal Processing and Communications, King Juan Carlos University, Madrid, Spain
A Graph Foundation Model for Wireless Resource Allocation
Apr 8, 2026Yucheng Sheng, Jiacheng Wang, Le Liang +2Wireless Resource Allocation Wireless Foundation Models
National Mobile Communications Research Laboratory, Southeast University, Nanjing 210096, China · Department of Electrical and Computer Engineering, University of California, Santa Cruz, CA 95064, USA
Limit Analysis of Graph Neural Networks with Wireless Conflict Graphs
Jun 2, 2026Romina Garcia Camargo, Zhiyang Wang, Alejandro RibeiroGraph Neural Networks Wireless Resource Allocation
Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia, PA · Halıcıoğlu Data Science Institute, UCSD, La Jolla, CA

Jul 7, 2026J/K move · Enter open · S save

Generative Diffusion Models of Stochastic Graph Signals

Yiğit Berkay Uslu, Samar Hadou, Sergio Rozada, Shirin Saeedi Bidokhti+1

Dept. of Electrical and Systems Eng., University of Pennsylvania, Philadelphia, PA 19104 USA · Department of Signal Processing and Communications, King Juan Carlos University, Madrid, Spain

Graph Neural Networks Financial Forecasting Conditional Diffusion Models Graph Diffusion

Open paper AskPDF arXiv

Sampling stochastic signals supported on a graph underlies many graph machine learning tasks, including recommender systems, forecasting in financial markets, and wireless network optimization. In these settings, the target signals are realizations of unknown conditional distributions. However, prevailing approaches rely mostly on intricate, application-tailored designs that often regress to a conditional mean instead of sampling from the conditional law. This paper unifies such problems as conditional graph signal generative modeling and tackles them with a single denoising diffusion framework. We learn a reverse diffusion process, parametrized by graph neural networks (GNNs), that draws graph signals conditioned directly on the graph topology and on node-feature side information. The reverse process is realized by a novel architecture, the U-Graph Neural Network (U-GNN), which generalizes the image-convolutional U-Net to graph-structured signals. The U-GNN performs multi-resolution encoder--decoder processing in which pooling and unpooling reduce to a learned node selection, expressed by nested selection matrices, and a zero-padded lifting of coarse signals back to the full node set. The graph convolutions are carried out on the original graph, with a stride that sets their hop reach, so the U-GNN bypasses explicit graph coarsening at every resolution. We demonstrate our method on two generative tasks: stock price forecasting and optimal wireless resource allocation, with extensive numerical results in both domains.

Abstract

Wireless Resource Allocation

Explore similar work

Generative Diffusion Models of Stochastic Graph Signals

A Graph Foundation Model for Wireless Resource Allocation

Limit Analysis of Graph Neural Networks with Wireless Conflict Graphs