Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Learning Based Channel Extrapolation for Dual-Band Massive MIMO Systems

Qikai Xiao, Kehui Li, Binggui Zhou, Shaodan Ma

TL;DR

The paper addresses the challenge of obtaining accurate mmWave CSI for dual-band massive MIMO while minimizing pilot overhead. It introduces the Multi-Domain Fusion Channel Extrapolator (MDFCE), a deep learning architecture that merges mixture-of-experts gating and multi-head self-attention to fuse spatial, frequency, and temporal features from sub-6 GHz CSI, enabling cross-band extrapolation to mmWave CSI. MDFCE comprises three modules—Temporal Feature Extraction Module (TFEM), Multi-Domain Fusion Module (MDFM), and Deep Feature Interaction Module (DFIM)—and optimizes a loss combining normalized MSE with a load-balancing auxiliary term to prevent expert collapse. Extensive experiments on the DeepMIMO dataset show MDFCE outperforms traditional pilot-based estimation under the same overhead, reduces pilot requirements by leveraging sub-6 GHz information, and offers significant computational efficiency gains compared to Transformer baselines, highlighting its potential for practical cross-band mmWave CSI acquisition.

Abstract

Future wireless communication systems will increasingly rely on the integration of millimeter wave (mmWave) and sub-6 GHz bands to meet heterogeneous demands on high-speed data transmission and extensive coverage. To fully exploit the benefits of mmWave bands in massive multiple-input multiple-output (MIMO) systems, highly accurate channel state information (CSI) is required. However, directly estimating the mmWave channel demands substantial pilot overhead due to the large CSI dimension and low signal-to-noise ratio (SNR) led by severe path loss and blockage attenuation. In this paper, we propose an efficient \textbf{M}ulti-\textbf{D}omain \textbf{F}usion \textbf{C}hannel \textbf{E}xtrapolator (MDFCE) to extrapolate sub-6 GHz band CSI to mmWave band CSI, so as to reduce the pilot overhead for mmWave CSI acquisition in dual band massive MIMO systems. Unlike traditional channel extrapolation methods based on mathematical modeling, the proposed MDFCE combines the mixture-of-experts framework and the multi-head self-attention mechanism to fuse multi-domain features of sub-6 GHz CSI, aiming to characterize the mapping from sub-6 GHz CSI to mmWave CSI effectively and efficiently. The simulation results demonstrate that MDFCE can achieve superior performance with less training pilots compared with existing methods across various antenna array scales and signal-to-noise ratio levels while showing a much higher computational efficiency.

Deep Learning Based Channel Extrapolation for Dual-Band Massive MIMO Systems

TL;DR

The paper addresses the challenge of obtaining accurate mmWave CSI for dual-band massive MIMO while minimizing pilot overhead. It introduces the Multi-Domain Fusion Channel Extrapolator (MDFCE), a deep learning architecture that merges mixture-of-experts gating and multi-head self-attention to fuse spatial, frequency, and temporal features from sub-6 GHz CSI, enabling cross-band extrapolation to mmWave CSI. MDFCE comprises three modules—Temporal Feature Extraction Module (TFEM), Multi-Domain Fusion Module (MDFM), and Deep Feature Interaction Module (DFIM)—and optimizes a loss combining normalized MSE with a load-balancing auxiliary term to prevent expert collapse. Extensive experiments on the DeepMIMO dataset show MDFCE outperforms traditional pilot-based estimation under the same overhead, reduces pilot requirements by leveraging sub-6 GHz information, and offers significant computational efficiency gains compared to Transformer baselines, highlighting its potential for practical cross-band mmWave CSI acquisition.

Abstract

Future wireless communication systems will increasingly rely on the integration of millimeter wave (mmWave) and sub-6 GHz bands to meet heterogeneous demands on high-speed data transmission and extensive coverage. To fully exploit the benefits of mmWave bands in massive multiple-input multiple-output (MIMO) systems, highly accurate channel state information (CSI) is required. However, directly estimating the mmWave channel demands substantial pilot overhead due to the large CSI dimension and low signal-to-noise ratio (SNR) led by severe path loss and blockage attenuation. In this paper, we propose an efficient \textbf{M}ulti-\textbf{D}omain \textbf{F}usion \textbf{C}hannel \textbf{E}xtrapolator (MDFCE) to extrapolate sub-6 GHz band CSI to mmWave band CSI, so as to reduce the pilot overhead for mmWave CSI acquisition in dual band massive MIMO systems. Unlike traditional channel extrapolation methods based on mathematical modeling, the proposed MDFCE combines the mixture-of-experts framework and the multi-head self-attention mechanism to fuse multi-domain features of sub-6 GHz CSI, aiming to characterize the mapping from sub-6 GHz CSI to mmWave CSI effectively and efficiently. The simulation results demonstrate that MDFCE can achieve superior performance with less training pilots compared with existing methods across various antenna array scales and signal-to-noise ratio levels while showing a much higher computational efficiency.
Paper Structure (10 sections, 20 equations, 6 figures, 2 tables)

This paper contains 10 sections, 20 equations, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. System Model and Problem Formulation
  3. System Model
  4. Problem Formulation
  5. Multi-Domain Fusion Channel Extrapolator
  6. Experiments
  7. Experimental Settings
  8. Effectiveness of the Cross-Band Channel Extrapolation
  9. Effectiveness of the MDFCE
  10. Conclusion

Figures (6)

  • Figure 1: The dual-band massive MIMO system.
  • Figure 2: Architecture of the proposed MDFCE.
  • Figure 3: Illustration of key components in the MDFCE.
  • Figure 4: Top view of selected locations in the scenario O1 of the DeepMIMO dataset alkhateeb2019deepmimo.
  • Figure 5: Comparison of pilot-based direct mmWave channel estimation and cross-band channel extrapolation.
  • ...and 1 more figures