Diversity combining and eigencombining performance and complexity comparison for estimated channels

Conventionally, antenna arrays apply maximal-ratio combining (MRC) directly to received signals. However, the performance of MRC compensates for its high numerical complexity only for low antenna correlation, i.e., rarely, for space-limited base-stations in actual scenarios, where the azimuth spread (AS) is random and predominantly small. Maximal-ratio eigencombining (MREC), i.e., MRC of the outputs of a partial Karhunen-Loeve Transform (KLT) of the received signals, can attain available diversity gain more efficiently. Herein, MREC is adapted to the AS with a bias-variance trade-off criterion (BVTC) that selects for KLT dominant eigenvectors of the channel correlation matrix that are computed with a deflation-based projection-approximation subspace tracking (PASTd) algorithm. Optimum and suboptimum channel fading estimation and weight-signal combining are evaluated. Simulations indicate that PASTd-BVTC-based suboptimum MREC with optimum fading estimation can perform similarly to the much more complex optimum-MRC approach.