[Paper Review] Estimating Doubly-Selective Channels for Hybrid mmWave Massive MIMO Systems: A Doubly-Sparse Approach

Shijian Gao, Xiang Cheng+1 more|arXiv (Cornell University)|Dec 18, 2019

Millimeter-Wave Propagation and Modeling25 references4 citations

TL;DR

This paper proposes a doubly-sparse channel estimation framework for hybrid mmWave massive MIMO systems that jointly exploits beamspace sparsity and delay-domain sparsity via a two-stage training design. By leveraging adaptive OMP (A-BOMP) and beamforming polling, the method achieves robust performance under double selectivity (frequency and time), reducing training overhead by 20% and memory usage by 99.9% compared to baseline methods, while maintaining high accuracy even in high-mobility scenarios.

ABSTRACT

In mmWave massive multiple-input multiple-output (mMIMO) systems, hybrid digital/analog beamforming has been recognized as an economic means to overcome the severe mmWave propagation loss. To facilitate beamforming for mmWace mMIMO, there is a great urgency to acquire accurate channel state information. To this end, a novel doubly-sparse approach is proposed to estimate doubly-selective mmWave channels under hybrid mMIMO. Via the judiciously designed training pattern, the well-known beamspace sparsity along with the under-investigated delay-domain sparsity that mmWave channels exhibit can be jointly exploited to assist channel estimation. Thanks to our careful two-stage (random-probing and steering-probing) design, the proposed channel estimator possesses strong robustness against the double (frequency and time) selectivity whilst enjoying the benefits brought by the exploitation of double sparsity. Compared with existing alternatives, our proposed channel estimator not only proves to be more general, but also largely reduces the training overhead, storage demand as well as computational complexity.

Motivation & Objective

• Address the challenge of accurate channel state information (CSI) acquisition in hybrid mmWave massive MIMO systems with limited RF chains.
• Overcome the limitations of existing methods that fail to exploit both beamspace and delay-domain sparsity simultaneously.
• Develop a robust channel estimator resilient to double selectivity (frequency and time variations) due to Doppler shifts in mobile environments.
• Minimize training overhead, storage demand, and computational complexity in wideband mmWave systems.
• Provide a general, efficient, and practical solution suitable for real-time implementation in 5G and beyond systems.

Proposed method

• Design a two-stage training pattern: random-probing to identify active channel taps and steering-probing to refine angle support using beamforming.
• Utilize energy detection to identify effective channel taps in the delay domain, enabling sparse representation of the wideband channel.
• Propose an adaptive OMP algorithm (A-BOMP) that dynamically adjusts basis matching and residue update to improve angle support recovery under high Doppler.
• Implement a beamforming polling strategy based on estimated angle supports to jointly estimate path gains and Doppler shifts with low training overhead.
• Leverage compressed sensing theory to exploit joint beamspace and delay-domain sparsity, reducing the number of required training symbols.
• Integrate Doppler compensation into the estimation process to maintain accuracy over time-varying channels.

Experimental results

Research questions

• RQ1 How can beamspace sparsity and delay-domain sparsity be jointly exploited to improve channel estimation in hybrid mmWave massive MIMO systems?
• RQ2 What training pattern design enables robust estimation under both frequency and time selectivity caused by Doppler shifts?
• RQ3 How does the proposed A-BOMP algorithm improve angle support recovery compared to conventional OMP or DPC-BOMP in high-mobility scenarios?
• RQ4 To what extent does the proposed doubly-sparse approach reduce training overhead and computational complexity compared to existing LS and BSA-based estimators?
• RQ5 How sensitive is the performance of the proposed method to the resolution of the analog phase shifter (APS) and the number of RF chains?

Key findings

• The proposed doubly-sparse approach (DSA) reduces training overhead by 20% compared to the baseline BSA at high SNR, demonstrating significant efficiency gains.
• DSA reduces memory usage from 18GB (BSA) to 9MB, a 99.9% reduction, due to a much smaller sensing matrix size of 200×4096.
• DSA outperforms BSA and LS estimators in NMSE across all SNR regions, with a notable performance gap at medium-to-high SNR.
• The A-BOMP algorithm achieves NMSE performance close to the ideal benchmark (K=4) even with pre-determined iterations, showing robustness and efficiency.
• The method maintains low NMSE degradation across varying numbers of propagation paths (P=3 to P=6), indicating insensitivity to path count.
• Increasing APS resolution from 1-bit to 2-bit improves NMSE by a large margin, but further gains from 3-5 bits are minimal (only 0.5dB improvement), showing insensitivity to fine resolution.