Capacity of MIMO Systems Aided by Microwave Linear Analog Computers (MiLACs)

TL;DR

This paper explores the fundamental capacity limits of MIMO systems enhanced by microwave linear analog computers (MiLACs), demonstrating that MiLACs can match digital beamforming capacity while reducing hardware complexity.

Contribution

It provides a closed-form solution for optimal MiLAC network design and capacity characterization, enabling scalable gigantic MIMO systems with reduced hardware costs.

Findings

MiLAC-aided beamforming achieves capacity equivalent to digital beamforming.

MiLACs significantly reduce RF chains and ADC/DAC resolution needs.

Theoretical results are validated by numerical simulations.

Abstract

Future wireless systems, known as gigantic multiple-input multiple-output (MIMO), are expected to enhance performance by significantly increasing the number of antennas, e.g., a few thousands. To enable gigantic MIMO overcoming the scalability limitations of digital architectures, microwave linear analog computers (MiLACs) have recently emerged. A MiLAC is a multiport microwave network that processes input microwave signals entirely in the analog domain, thereby reducing hardware costs and computational complexity of gigantic MIMO architectures. In this paper, we investigate the fundamental limits on the rate achievable in MiLAC-aided MIMO systems. We model a MIMO system employing MiLAC-aided beamforming at the transmitter and receiver, and formulate the rate maximization problem to optimize the microwave networks of the MiLACs, which are assumed lossless and reciprocal for practical reasons. Under the lossless and reciprocal constraints, we derive a global optimal solution for the microwave networks of the MiLACs in closed form. In addition, we also characterize in closed-form the capacity of MIMO systems operating MiLAC-aided beamforming. Our theoretical analysis, confirmed by numerical simulations, reveals that MiLAC-aided beamforming achieves the same capacity as digital beamforming, while significantly reducing the number of radio frequency (RF) chains, analog-to-digital converters (ADCs)/digital-to-analog converters (DACs) resolution requirements, and computational complexity.