Semi-blind space-time estimation of co-channel signals using least squares projections

We propose a new method, joint space-time DWILSF (decoupled weighted iterative least squares with subspace fitting) algorithm that maps array output signals to a known finite alphabet (FA) symbol constellation without preceding subspace estimation. We evaluate the performance of our approach by simulating space division multiple access (SDMA) and spatial filtering for interference reduction (SFIR) scenarios using the GSM radio interface. Two different semi-blind receivers act as performance references. We show that our linear approach requiring only matrix multiplications during iterations converges fast and provides promising performance with reasonable computational complexity