Near-Optimum Soft-Output Ant-Colony-Optimization Based Multiuser Detection for the DS-CDMA Uplink

In this contribution, a novel soft-output Ant Colony Optimization (ACO) based Multi-User Detector (MUD) is proposed for the synchronous Direct-Sequence Code-Division- Multiple-Access (DS-CDMA) uplink. The foraging behaviour of the ant colony in nature motivates the employment of reduced- search ACO-based MUDs, which are capable of approaching the optimum Maximum Likelihood (ML) MUD's performance at the cost of a computational complexity, which maybe as low as that of the Mathched Filter (MF) based Single User Detector (SUD). However, the previously proposed conventional ACO based MUDs were unable to provide soft Log-Likelihood Ratio (LLR) values for the channel decoder. Hence in this paper, we present a novel soft-output ACO-MUD capable of delivering soft LLRs, which allows a CDMA system to achieve a near-single-user performance without any additional information feedback from the channel decoder, even when the number of users supported is as high as the number of chip in the spreading sequence. Our numerical results show that at a BER of 10
^-3
, the performance of the currently known ACO-assisted state-of-the-art systems can be improved by about 17 dB with the aid of the proposed soft-output ACO-MUD. More explicitly, the soft-output ACO- MUD is capable of approaching the optimum performance of the Bayesian detector, when K = 32 UL users are supported with the aid of 31-chip Gold codes, while the complexity of the former is only a fraction of 10
^-8
of the latter.