Dual-Attention Based 3D Channel Estimation

2026-04-02

AI Summary

This work addresses the high computational complexity of conventional three-dimensional (3D) channel estimation in massive MIMO systems, which stems from high-dimensional matrix operations, and the significant performance degradation of existing suboptimal methods under strongly correlated channels. To overcome these limitations, the paper proposes a novel 3D Channel Estimation Network (3DCENet) that, for the first time, integrates a dual attention mechanism into this task to jointly model multidimensional channel correlations across the time, frequency, and spatial domains. The proposed method achieves estimation accuracy approaching the linear minimum mean square error (LMMSE) theoretical bound while substantially reducing computational complexity, and it demonstrates markedly superior performance over current suboptimal approaches—particularly in scenarios with strong channel correlation.

Abstract

For multi-input and multi-output (MIMO) channels, the optimal channel estimation (CE) based on linear minimum mean square error (LMMSE) requires three-dimensional (3D) filtering. However, the complexity is often prohibitive due to large matrix dimensions. Suboptimal estimators approximate 3DCE by decomposing it into time, frequency, and spatial domains, while yields noticeable performance degradation under correlated MIMO channels. On the other hand, recent advances in deep learning (DL) can explore channel correlations in all domains via attention mechanisms. Building on this capability, we propose a dual attention mechanism based 3DCE network (3DCENet) that can achieve accurate estimates.

Problem

• 3D channel estimation
• MIMO channels
• LMMSE
• channel correlation
• computational complexity

Innovation

• dual attention
• 3D channel estimation
• MIMO
• deep learning
• channel correlation