ML Detection for MIMO Systems under Channel Estimation Errors

TL;DR

This paper introduces a robust ML detection scheme for MIMO systems that accounts for channel estimation errors, improving error performance through an optimized covariance estimation and an iterative decoding approach.

Contribution

It presents a novel ML detection method that is resilient to channel estimation errors and includes an iterative sphere decoding algorithm to reduce complexity.

Findings

Performance gain over existing algorithms in error probability

Effective covariance estimation under bounded channel errors

Reduced computational complexity with iterative decoding

Abstract

In wireless communication systems, the use of multiple antennas at both the transmitter and receiver is a widely known method for improving both reliability and data rates, as it increases the former through transmit or receive diversity and the latter by spatial multiplexing. In order to detect signals, channel state information (CSI) is typically required at the receiver; however, the estimation of CSI is not perfect in practical systems, which causes performance degradation. In this paper, we propose a novel maximum likelihood (ML) scheme that is robust to channel information errors. By assuming a bound on the total power of channel estimation errors, we apply an optimization method to estimate the instantaneous covariance of channel estimation errors in order to minimize the ML cost function. To reduce computational complexity, we also propose an iterative sphere decoding scheme based on the proposed ML detection method. Simulation results show that the proposed algorithm provides a performance gain in terms of error probability relative to existing algorithms.