Low Complexity Turbo Detection of Coded Under-Determined MIMO Systems

Abstract

The turbo detection of a coded under-determined multiple-input multiple-output (UD-MIMO) system is studied in this paper. A UD-MIMO system with $N$ transmit antennas and $M < N$ receive antennas has more unknowns than the observations. However, most existing low complexity detectors, such as the vertical Bell Laboratories Layered Space-Time (V-BLAST) or the block decision feedback equalizer (BDFE), work only when $M \geq N$. We propose a new turbo detector structure that combines the existing generalized parallel interference cancellation (GPIC) technique and a new generalized serial interference cancellation (GSIC) technique, with a soft-input soft-output (SISO) BDFE. In the first iteration, the GPIC-BDFE equalization is adopted. From the second iteration and beyond, due to the availability of the a priori soft information, the UD-MIMO system can be partitioned into multiple sub-systems by using a reliability-based partition scheme. Then, the GSIC-BDFE equalization, which has a much lower complexity than the GPIC-BDFE equalization, can be used. Simulation results show that the new detection scheme achieves significant performance gain as the iteration progresses with a reasonable complexity.