Rate maximization in coupled MIMO systems: A generic algorithm for designing single-port matching networks

Compact arrays are known to be associated with antenna coupling and noise correlation. The noise can be either antenna noise, LNA noise or downstream noise. Due to these effects, it was shown that the matching network affects the performance of MIMO systems with coupled receiver antennas. Since the optimal multiport matching network is of very high complexity as well as very narrow operation bandwidth, development of single-port (SP) matching networks that boost the performance became inevitable. In this paper we develop a gradient-search algorithm to design the matching network for achievable rate maximization of multi user MIMO systems. For any combination of noise sources, we rigorously derive the exact gradient of the achievable rate with respect to the components of the matching network. We assume either full knowledge of the spatial channel or knowing its statistical properties. In the later case we optimize the matching network to maximize the Jensen's bound. Substantial performance enhancement is shown when our algorithms are used. Significant reduction in the array area is gained in comparison to the often used λ/2 antenna spacing without taking coupling into account. This can be vital for future wireless systems adopting massive MIMO arrays. Via eigenvalues distribution simulations at different SNR regimes, we show an intuitive link to the communication theory.