A LOW COMPLEXITY SQUARE ROOT MMSE MIMO DECODER

As MIMO technology proliferates in the latest wireless communication systems, efficient architectures for MIMO Decoding are being pursued. An alternative formulation of the MMSE MIMO Decoder for spatial multiplexing MIMO systems called the square root MMSE MIMO Decoder based on QR decomposition has appeared in the literature. It avoids inverting the upper triangular matrix but at the cost of dealing with a larger matrix, the extended channel matrix. In this paper, we present an approach based on the square root MMSE scheme where we don't pay the price for the extended channel matrix in terms of increased hardware resources but at the same time benefit from its better fixed precision properties thereby achieving a solution with much lower complexity. We present a scalable, systolic array based solution where the systolic array scales with the number of transmit antennas and not with the number of receive antennas. Hence, the complexity of the MIMO Decoder does not increase even as the number of antennas at the receiver (base-station) is increased.