Variable List Detection for Multiuser MIMO Systems

In this paper, we propose a list detection scheme for the uplink scenario of multiuser multiple-input-multiple-output (MU-MIMO) systems. The proposed technique employs a single lattice reduction (LR) transformation to modify the channel matrix between the users and the base station, which produces a new nearly orthogonal channel matrix and allows for a more efficient detection. After the LR transformation, a reliable candidate for the transmitted signal vector, which is provided by successive interference cancelation detection with a cancelation ordering strategy based on the mean square error in the LR domain, is obtained. An algorithm to decide if the current candidate has good quality or if it is necessary to further explore different orderings to improve the estimation quality is employed. The proposed algorithm sets a minimum number of candidates to be tested and employs a stopping criterion to keep the number of candidates from growing excessively. Simulation results indicate that the proposed variable list detection (VLD) scheme has a near-optimal performance while keeping its computational complexity well below that of the maximum-likelihood detector. An iterative detection and decoding scheme based on the VLD algorithm is also developed, producing an excellent performance that approaches the single-user scenario.