Efficient Soft-Input Soft-Output MIMO Detection via Improved M-Algorithm

In this paper, we propose a new soft-input soft-output (SISO) multi-input multi-output (MIMO) detection technique, called an improved SISO M-algorithm (ISS-MA). We modify the conventional M-algorithm to improve the performance-complexity trade-off of the SISO symbol detector. Towards this end, an improved path metric is proposed, which accounts for the information on undecided symbols at a particular path visited. The inclusion of this information is enabled through a bias term which is added to the conventional path metric in order to reflect the contributions of the undecided symbols. We derive the bias term using soft unconstrained linear estimates of undecided symbols. As a result, the ISS-MA that picks up the best M candidates based on this modified path metric exhibits improved performance/complexity trade-off compared to the existing SISO detectors. According to extensive simulations performed over i.i.d. Rayleigh fading channels, the proposed SISO detector yields significantly lower complexity than other symbol detectors while maintaining strong performance especially in high dimensional systems.