Co-learning-aided Multi-modal-deep-learning Framework of Passive DOA Estimators for a Heterogeneous Hybrid Massive MIMO Receiver

TL;DR

This work tackles DOA estimation in massive MIMO by addressing HAD-induced phase ambiguity and latency with a novel heterogeneous H^2AD-FD MIMO receiver. It introduces a three-stage multi-modal learning framework that combines Root-MUSIC and CNN-based coarse DOA estimates, ML-based true-solution inference, and fusion to yield high-accuracy DOA estimates, supported by CRLB analysis. The paper presents four estimators—MD-Root-MUSIC, MDDL, CoMDDL, and CoMD-RootMUSIC—and derives closed-form fusion weights and CRLBs, demonstrating near-CRLB performance at moderate to high SNR and significant gains in low-SNR regimes via co-learning. Overall, the H^2AD-FD architecture reduces cost and latency while maintaining accuracy, offering a promising path for efficient, green wireless sensing and ISAC deployments.

Abstract

Due to its excellent performance in rate and resolution, fully-digital (FD) massive multiple-input multiple-output (MIMO) antenna arrays has been widely applied in data transmission and direction of arrival (DOA) measurements, etc. But it confronts with two main challenges: high computational complexity and circuit cost. The two problems may be addressed well by hybrid analog-digital (HAD) structure. But there exists the problem of phase ambiguity for HAD, which leads to its low-efficiency or high-latency. Does exist there such a MIMO structure of owning low-cost, low-complexity and high time efficiency at the same time. To satisfy the three properties, a novel heterogeneous hybrid MIMO receiver structure of integrating FD and heterogeneous HAD ($\rm{H}^2$AD-FD) is proposed and corresponding multi-modal (MD)-learning framework is developed. The framework includes three major stages: 1) generate the candidate sets via root multiple signal classification (Root-MUSIC) or deep learning (DL); 2) infer the class of true solutions from candidate sets using machine learning (ML) methods; 3) fuse the two-part true solutions to achieve a better DOA estimation. The above process form two methods named MD-Root-MUSIC and MDDL. To improve DOA estimation accuracy and reduce the clustering complexity, a co-learning-aided MD framework is proposed to form two enhanced methods named CoMDDL and CoMD-RootMUSIC. Moreover, the Cramer-Rao lower bound (CRLB) for the proposed $\rm{H}^2$AD-FD structure is also derived. Experimental results demonstrate that our proposed four methods could approach the CRLB for signal-to-noise ratio (SNR) > 0 dB and the proposed CoMDDL and MDDL perform better than CoMD-RootMUSIC and MD-RootMUSIC, particularly in the extremely low SNR region.

Figures (10)

• Figure 1: Proposed heterogeneous hybird array with FD and $\rm{H}^2$AD, where $\rm{H}^2$AD has $H$ groups and group $h$ has $K_h$ subarrays with each subarray having $M_h$ antennas, $h\in\{1,2,\cdots,H\}$
• Figure 2: Multi-modal framework for estimating DOA.
• Figure 3: Co-learning-aided multi-modal framework for estimating DOA.
• Figure 4: The poposed CNN--based DOA estimation for FD Array.
• Figure 5: The training and validation loss of the CNN.