# Machine Learning for Beam Alignment in Millimeter Wave Massive MIMO

arXiv: https://arxiv.org/abs/2002.06362 · HTML: https://mlsanity.com/paper/2002.06362/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 15 Feb 2020.

## Built on by (most cited first, 4 of 4)

- [Graph Attention Networks for Channel Estimation in RIS-assisted Satellite IoT Communications](https://mlsanity.com/paper/2104.00735/) (2021)
- [On Single-User Interactive Beam Alignment in Next Generation Systems: A Deep Learning Viewpoint](https://mlsanity.com/paper/2102.10229/) (2021)
- [Graph Attention Network Based Single-Pixel Compressive Direction of Arrival Estimation](https://mlsanity.com/paper/2109.05466/) (2021)
- [AI-Driven Mobility Management for High-Speed Railway Communications: Compressed Measurements and Proactive Handover](https://mlsanity.com/paper/2407.04336/) (2024)

## Abstract (arXiv)

This article investigates beam alignment for multi-user millimeter wave (mmWave) massive multi-input multi-output system. Unlike the existing works using machine learning (ML), an alignment method with partial beams using ML (AMPBML) is proposed without any prior knowledge such as user location information. The neural network (NN) for the AMPBML is trained offline using simulated environments according to the mmWave channel model and is then deployed online to predict the beam distribution vector using partial beams. Afterwards, the beams for all users are all aligned simultaneously based on the indices of the dominant entries of the obtained beam distribution vector. Simulation results demonstrate that the AMPBML outperforms the existing methods, including the adaptive compressed sensing, hierarchical search, and multi-path decomposition and recovery, in terms of the total training time slots and the spectral efficiency.