Statistics-based antenna selection for multiaccess MIMO systems

Receive antenna selection based on statistical channel knowledge (SCK) is considered for a multiuser uplink system. For spatial multiplexing transmission, we derive the optimal SCK-based antenna selection criteria that maximizes the ergodic capacity in the high signal to noise ratio regime. Two different receiver architectures are considered. The first receiver uses a MMSE front-end to suppress the interference, followed by ML decoding to optimally decode the desired user's data streams and the second receiver performs a joint ML detection of all the users' data streams. In a sufficiently correlated channel, SCK-based selection gives gains close to exact channel knowledge-based selection. We compare the performance of antenna selection with that of full complexity (FC) receiver, that uses all the available antennas for demodulation. We show that, unlike in the single-user system, in a multiaccess system, the capacity achieved by a MMSE receiver with antenna selection can be higher than that of a MMSE FC receiver.