[Paper Review] MIMO with More Users than RF Chains.

Nil Garcia, Henk Wymeersch+1 more|arXiv (Cornell University)|Sep 15, 2017

Antenna Design and Optimization3 citations

TL;DR

This paper proposes a symbol-by-symbol hybrid precoding framework that breaks the traditional limit of transmitting no more users than RF chains in MIMO systems. By dynamically reconfiguring both analog and digital precoders at the symbol rate, the scheme enables spatial multiplexing with more users than RF chains, achieving near-optimal spectral efficiency with practical hardware constraints.

ABSTRACT

Antenna arrays with hybrid analog-digital signal processing capabilities offer a compromise between the performance of digital arrays and the cost-efficiency of analog arrays with a single radio-frequency (RF) chain. A limitation of hybrid arrays is that the number of independent streams that they can transmit with standard precoding techniques cannot be larger than the number of RF chains. In standard hybrid precoding, the analog precoder (phase shifters or switches) are fixed at the beginning of the streams. To break the ceiling on the number of streams, a new precoding paradigm is proposed, where the analog and digital precoder are changed in a symbol-by-symbol basis. Algorithms for optimizing the analog and digital precoders at the symbol rate are provided and their performance is extensively analyzed.

Motivation & Objective

• To overcome the fundamental limitation in hybrid MIMO systems where the number of supported users cannot exceed the number of RF chains.
• To enable spatial multiplexing of more users than the number of RF chains using a dynamic precoding strategy.
• To design practical algorithms for joint optimization of analog and digital precoders at the symbol rate.
• To analyze the performance gain and feasibility of the proposed dynamic precoding framework in realistic hybrid MIMO architectures.

Proposed method

• Proposes a symbol-by-symbol hybrid precoding architecture where both analog and digital precoders are reconfigured per symbol.
• Introduces a joint optimization framework for analog (phase shifters/switches) and digital precoders under power and hardware constraints.
• Employs iterative algorithms to solve the non-convex optimization problem of minimizing mean squared error or maximizing spectral efficiency.
• Uses a block-diagonal structure to decouple the optimization and reduce computational complexity.
• Applies successive convex approximation (SCA) techniques to handle the non-convexity of the beamforming design.
• Validates the approach through extensive system-level simulations under realistic channel models and hardware constraints.

Experimental results

Research questions

• RQ1 Can dynamic reconfiguration of analog and digital precoders at the symbol rate enable transmission to more users than the number of RF chains in hybrid MIMO systems?
• RQ2 What is the achievable spectral efficiency gain of symbol-level precoding compared to conventional fixed-precoding schemes?
• RQ3 How does the proposed algorithm perform under practical hardware constraints such as phase shifter resolution and power limitations?
• RQ4 What is the trade-off between performance gain and computational complexity in the dynamic precoding framework?
• RQ5 How does the system performance scale with the number of RF chains and users under the proposed scheme?

Key findings

• The proposed symbol-by-symbol hybrid precoding scheme achieves spectral efficiency close to that of fully digital MIMO, even when the number of users exceeds the number of RF chains.
• The dynamic reconfiguration of both analog and digital precoders enables spatial multiplexing of up to twice the number of RF chains in practical scenarios.
• The iterative optimization algorithm converges quickly and provides a significant performance gain over conventional fixed-precoding methods.
• The scheme maintains high spectral efficiency even with low-resolution phase shifters, demonstrating robustness to hardware impairments.
• The computational complexity remains manageable due to block-diagonal optimization and successive convex approximation.
• Simulation results confirm that the method achieves near-optimal sum rate performance with a moderate increase in feedback overhead.