Likelihood ascent search augmented sphere decoding receiver for MIMO systems using M‐QAM constellations

Abstract

MIMO systems employing sphere decoding (SD) algorithm are known to achieve near maximum likelihood (ML) performance at a reduced complexity by restricting the candidate search space to a sphere of a certain radius. The performance of SD depends on the precise estimation of its soft output. In this paper, a low complexity modified Likelihood Ascent Search (LAS) algorithm is proposed to be used within a SD receiver in order to precisely estimate the counter-hypothesis for its winner candidates. The LAS algorithm is modified to search for the best counter-hypothesis in only one-half of the signal lattice thereby improving the performance of MIMO receiver. Our results challenge the popular perception that for a SD receiver a large number of candidates within the search sphere is essential for good performance. Instead, it is shown that accurate estimation of the counter-hypothesis is equally important and in fact, the performance of the proposed augmented SD receiver with only single candidate approaches that of a classical SD with multiple candidates. Bit error rate performance of the proposed method when compared with the existing research works on soft output generation for the same number of candidates shows that our proposed method outperforms them by upto 3 dB.

Soft-Output Deep-LAS Detection for Coded MIMO System: A Learning-Aided LLR Approximation

“…The MIMO detection problem based on the optimal ML detection algorithm is equivalent to minimize the Euclidean distance and solving the problem based on criterion [9] x = min…”

Soft-Output Deep-LAS Detection for Coded MIMO System: A Learning-Aided LLR Approximation

“…Recently, several optimal and sub-optimal MIMO detection algorithms have been proposed in the literature and can be broadly characterized into maximum likelihood (ML) detection [4]- [9], linear detectors [10], [11], interference cancellation based algorithms [12], [13], graph-based [14], [15], sparsity boosted [16]- [18], lattice-reduction-based [19]- [21] and neighborhoodsearch-based algorithms [22]- [28]. The optimal detector, such as ML and maximum a posteriori (MAP), recover all the symbols jointly and simultaneously.…”

“…The soft-output SD algorithm for MIMO and distributed antenna systems was proposed in [8], [9], where the authors showed that for the worst-case scenario the SD's performance could be improved by precise and low-complexity estimation of soft-output for SD. Nevertheless, the detection complexity of SD is still higher in high dimensional MIMO systems.…”