Joint Beamforming and Integer User Association using a GNN with Gumbel-Softmax Reparameterizations

TL;DR

This paper introduces a GNN-based approach with Gumbel-Softmax reparameterization for joint beamforming and integer user association in wireless networks, achieving higher sum-rate and exact integer solutions without added complexity.

Contribution

It presents a novel GNN architecture that guarantees integer user association using Gumbel-Softmax, improving over fractional methods in network optimization.

Findings

Achieves integer user association with Gumbel-Softmax reparameterization.

Outperforms fractional association methods in sum-rate performance.

Maintains computational efficiency while ensuring integer solutions.

Abstract

Machine learning (ML) models can effectively optimize a multi-cell wireless network by designing the beamforming vectors and association decisions. Existing ML designs, however, often needs to approximate the integer association variables with a probability distribution output. We propose a novel graph neural network (GNN) structure that jointly optimize beamforming vectors and user association while guaranteeing association output as integers. The integer association constraints are satisfied using the Gumbel-Softmax (GS) reparameterization, without increasing computational complexity. Simulation results demonstrate that our proposed GS-based GNN consistently achieves integer association decisions and yields a higher sum-rate, especially when generalized to larger networks, compared to all other fractional association methods.