Matrix-Lifting Semi-Definite Programming for Decoding in Multiple Antenna Systems

TL;DR

This paper introduces a matrix-lifting SDP-based decoder for multiple antenna systems that reduces computational complexity and improves performance over existing quasi-ML decoding methods, applicable to various constellations and labelings.

Contribution

It proposes a novel relaxation algorithm that significantly decreases the number of variables in SDP decoding for MIMO systems, enhancing efficiency and accuracy.

Findings

Reduces SDP variables from (NK+1)^2 to (2N+K)^2, lowering computational complexity.

Achieves better performance than existing quasi-ML decoding methods.

Applicable to any constellation and labeling method.

Abstract

This paper presents a computationally efficient decoder for multiple antenna systems. The proposed algorithm can be used for any constellation (QAM or PSK) and any labeling method. The decoder is based on matrix-lifting Semi-Definite Programming (SDP). The strength of the proposed method lies in a new relaxation algorithm applied to the method of Mobasher et al. This results in a reduction of the number of variables from $(NK+1)^2$ to $(2N+K)^2$, where $N$ is the number of antennas and $K$ is the number of constellation points in each real dimension. Since the computational complexity of solving SDP is a polynomial function of the number of variables, we have a significant complexity reduction. Moreover, the proposed method offers a better performance as compared to the best quasi-maximum likelihood decoding methods reported in the literature.