A nonlinear iterative beamforming technique for wireless communications

In this paper, we propose an iterative beamformer for multi-transmit, multi-receive wireless communications. The paper introduces an optimal MAP receiver with implementation complexity exponential in terms of the number of transmitting antennas, and a suboptimal nonlinear parallel interference cancelation scheme that significantly reduces the degradation effect of interference with implementation complexity linear in the number of transmitting antennas. The proposed receiver has a two-stage learning strategy: a linear beamforming phase using a short training sequence for learning the linear decision regions, and a nonlinear iterative scheme that makes use of the implicit form of supervision provided by a forward error-correction code (FEC) for tracking the nonlinear optimal decision boundaries. Simulations are provided as experimental evidence for the performance of the new receiver.