[Paper Review] Diversity Analysis of Millimeter-Wave Massive MIMO Systems

Dian‐Wu Yue, Shuai Xu+1 more|arXiv (Cornell University)|Jan 1, 2018

Advanced MIMO Systems Optimization25 references3 citations

TL;DR

This paper proposes a distributed antenna-subarray architecture for millimeter-wave massive MIMO systems to enhance diversity gain, showing that with infinitely many antennas per subarray, the diversity gain scales as $K_rK_tL - N_s + 1$, where $K_r$, $K_t$ are receiver and transmitter subarrays, $L$ is the number of propagation paths, and $N_s$ is the number of data streams. This architecture significantly outperforms co-located and conventional partially-connected structures in diversity performance under high SNR.

ABSTRACT

This paper is concerned with asymptotic diversity analysis for millimeter-wave (mmWave) massive MIMO systems. First, for a single-user mmWave system employing distributed antenna subarray architecture in which the transmitter and receiver consist of Kt and Kr subarrays, respectively, a diversity gain theorem is established when the numbers of antennas at subarrays go to infinity. Specifically, assuming that all subchannels have the same number of propagation paths L, the theorem states that by employing such a distributed antenna-subarray architecture, a diversity gain of KrKtL-Ns+1 can be achieved, where Ns is the number of data streams. This result means that compared to the co-located antenna architecture, using the distributed antenna-subarray architecture can scale up the diversity gain or multiplexing gain proportionally to KrKt. The diversity gain analysis is then extended to the multiuser scenario as well as the scenario with conventional partially-connected RF structure in the literature. Simulation results obtained with the hybrid analog/digital processing corroborate the analysis results and show that the distributed subarray architecture indeed yields significantly better diversity performance than the co-located antenna architectures.

Motivation & Objective

• To analyze the asymptotic diversity gain in mmWave massive MIMO systems under realistic propagation conditions.
• To investigate how distributed subarray architectures improve diversity and multiplexing gains compared to co-located or conventional partially-connected RF architectures.
• To establish a theoretical framework for diversity-multiplexing tradeoff (DMT) in mmWave massive MIMO with limited RF chains.
• To validate the theoretical findings through simulations using hybrid analog/digital precoding and combiner designs.

Proposed method

• Derives a diversity gain theorem for single-user mmWave systems with distributed subarrays, assuming infinite antennas per subarray and identical path counts $L$ across all subchannels.
• Uses singular value decomposition (SVD) of subchannel matrices and applies ZF digital precoding in multiuser scenarios to isolate user-specific channels.
• Analyzes the diversity performance under both homogeneous and inhomogeneous large-scale fading coefficient distributions $\{g_{ij}\}$.
• Employs hybrid analog/digital beamforming with fully connected and partially connected RF architectures for comparison.
• Introduces a generalized framework to compute diversity gain based on the number of subarrays $K_r$, $K_t$, propagation paths $L$, and data streams $N_s$.
• Validates theoretical results via Monte Carlo simulations with BER performance curves under varying SNR and system configurations.

Experimental results

Research questions

• RQ1 How does the diversity gain scale in mmWave massive MIMO systems when using distributed subarrays with infinite antennas per subarray?
• RQ2 What is the impact of the number of subarrays $K_r$ and $K_t$ on the achievable diversity gain compared to co-located or partially-connected architectures?
• RQ3 How does the diversity-multiplexing tradeoff (DMT) differ between distributed subarray and conventional RF architectures in mmWave systems?
• RQ4 Does inhomogeneous large-scale fading affect the diversity gain, and if so, how?
• RQ5 Can the proposed distributed subarray architecture achieve full diversity gain under practical hybrid precoding constraints?

Key findings

• The diversity gain for a single-user mmWave system with distributed subarrays is $K_rK_tL - N_s + 1$, which increases proportionally with the product $K_rK_t$.
• The conventional partially-connected RF architecture achieves only full diversity gain of 3 in the studied setup, while the distributed subarray architecture achieves up to 9 diversity gain with $K=2$ and $N_s=4$.
• In multiuser downlink scenarios, increasing the number of subarrays at the base station ($K_b$) significantly improves diversity performance, with gains increasing as $K_b$ increases.
• The diversity gain remains unchanged under inhomogeneous large-scale fading coefficient distributions, indicating robustness to path loss variations across subarrays.
• The hybridly-connected (overlapped subarray-based) structure achieves better spectral efficiency than conventional partially-connected structures and approaches fully-connected performance with sufficient RF chains.
• Simulations confirm that the distributed subarray architecture yields significantly better BER performance than co-located and conventional partially-connected architectures in the high SNR regime.