Divide-and-Conquer Matrix Inversion for Linear MMSE Detection in SDR MIMO Receivers

Software-defined radio (SDR) platforms represent a promising approach to facing the demands of the fast evolving field of wireless communications. The flexibility of SDR solutions is typically obtained at the cost of limited processing power, which imposes the use of low-complexity algorithms. This task is particularly challenging within multiple-input multiple-output (MIMO) communication, which is inherently characterized by heavy signal processing load. In this paper, we present a recursive matrix inversion algorithm for Hermitian positive-definite (HPD) matrices. The algorithm exhibits low computational complexity and is thus particularly suited for the HPD matrix inversion required in MIMO minimum mean squared error receivers. The implementation on a design-time configurable VLIW processor, configured with appropriate processing units and fabricated in 0.18 ¿m 1P/6M CMOS technology, demonstrates that real-time operation in IEEE 802.11n-like MIMO-OFDM systems with up to 52 carriers and 3 antennas at the transmitter and at the receiver is possible.