Evaluation of Optimum SIMO Combining for Estimated and Correlated Rician Fading

This analysis extends to Rician fading our earlier studies for Rayleigh fading of maximal-ratio eigencombining (MREC) — also known as eigenbeamforming — vs. maximal-ratio combining (MRC) and statistical beamforming (BF), for single-input multiple-output (SIMO) wireless systems. MREC is a superset of BF and MRC. Given instantaneous channel state information (ICSI), which is the channel gain vector estimate, and statistical CSI, which is determined by the Rician K-factor and the azimuth spread (AS), we analyze the optimum combining approach (instead of the commonly-studied suboptimal combining approach) based on the distribution of the effective signal-to-noise ratio. An expression is derived for the average error probability (AEP) of MREC in correlated Rician fading when antenna K-factors may be unequal. This AEP also applies for MRC and BF with estimated ICSI. Furthermore, a new bias–variance tradeoff criterion (BVTC) is proposed for MREC adaptation not only to AS (as with existing criteria), but also to K and ICSI accuracy. First, we demonstrate that BVTC-based adaptive MREC can achieve near-MRC AEP performance and lower numerical complexity. Furthermore, we find that the results of the realistic MREC performance and complexity evaluation approach, i.e., averaging over measured K distribution, are substantially different than the results of the conventional evaluation approach, i.e., setting K to typical value (distribution average or zero).