Receive Antenna Selection with Imperfect Channel Knowledge From Training

Receive antenna selection (AS), in which a subset of a receiver’s antennas, is selected based on the instantaneous channel condition, enables the receiver to benefit from having multiple antenna elements despite using significantly less hardware. However, the limited hardware resources also impose unique constraints on how channel state information, which is necessary to select the antennas and decode data, can be acquired for AS. We derive novel closed-form expressions for the symbol error probability of MPSK and MQAM over time-varying Rayleigh fading channels using AS. Our analysis explicitly considers a practical and general training model that accounts for noisy channel estimates and training delays that affect both selection and data decoding. Our results show that training plays a crucial role in determining the overall performance, and can even limit the number of receive antennas that can gainfully be employed in the system.