A Generalized MMSE Detection with Reduced Complexity for Spatially Multiplexed MIMO Signals

TL;DR

This paper introduces a low-complexity generalized MMSE detection method for spatially multiplexed MIMO systems, approximating the optimal detection with minimal performance loss and significantly reduced computational complexity.

Contribution

It proposes a novel approximation technique using uniform ring approximation for the probability density function, enabling near-MLD performance with linear complexity for two transmit antennas.

Findings

Performance comparable to MLD with reduced complexity

Uniform ring approximation effective for PSK and APSK

Linear complexity achieved when transmit antennas are two

Abstract

In multiple-input multiple-output (MIMO) spatially multiplexing (SM) systems, achievable error rate performance is determined by signal detection strategy. The optimal maximum-likelihood detection (MLD) that exhaustively examines all symbol candidates has exponential complexity and may not be applicable in many practical systems. In this paper, we consider a generalized minimum mean square error (MMSE) detection derived from conditional mean estimation, which in principle behaves equivalently to MLD but also includes a linear MMSE detection as a special case. Motivated by this fact, we propose a low-complexity detection which significantly reduces the number of examined symbol candidates without significant error rate performance degradation from MLD. Our approach is to approximate the probability density function (pdf) of modulated symbols that appears in the exact conditional mean expression such that the decision metric can be cast into a partially closed form. It is found that uniform ring approximation in combination with phase shift keying (PSK) and amplitude phase shift keying (APSK) is promising, as it can achieve a performance even comparable to MLD, while its complexity is linear when the number of transmit antennas is two.