# Edge Graph Neural Networks for Massive MIMO Detection

arXiv: https://arxiv.org/abs/2206.06979 · HTML: https://mlsanity.com/paper/2206.06979/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 18 references that are in the map.
- First posted on arXiv 22 May 2022.

## Builds on (most cited first, 7 of 7)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Semi-Supervised Classification with Graph Convolutional Networks](https://mlsanity.com/paper/1609.02907/) (2016)
- [Graph Attention Networks](https://mlsanity.com/paper/1710.10903/) (2017)
- [Gated Graph Sequence Neural Networks](https://mlsanity.com/paper/1511.05493/) (2015)
- [Inference in Probabilistic Graphical Models by Graph Neural Networks](https://mlsanity.com/paper/1803.07710/) (2018)
- [Graph Neural Networks for Massive MIMO Detection](https://mlsanity.com/paper/2007.05703/) (2020)
- [A Low-Complexity ADMM-based Massive MIMO Detectors via Deep Neural Networks](https://mlsanity.com/paper/2103.00131/) (2021)

## Abstract (arXiv)

Massive Multiple-Input Multiple-Out (MIMO) detection is an important problem in modern wireless communication systems. While traditional Belief Propagation (BP) detectors perform poorly on loopy graphs, the recent Graph Neural Networks (GNNs)-based method can overcome the drawbacks of BP and achieve superior performance. Nevertheless, direct use of GNN ignores the importance of edge attributes and suffers from high computation overhead using a fully connected graph structure. In this paper, we propose an efficient GNN-inspired algorithm, called the Edge Graph Neural Network (EGNN), to detect MIMO signals. We first compute graph edge weights through channel correlation and then leverage the obtained weights as a metric to evaluate the importance of neighbors of each node. Moreover, we design an adaptive Edge Drop (ED) scheme to sparsify the graph such that computational cost can be significantly reduced. Experimental results demonstrate that our proposed EGNN achieves better or comparable performance to popular MIMO detection methods for different modulation schemes and costs the least detection time compared to GNN-based approaches.