Predictive receive-directed antenna selection for quasistatic Rayleigh fading channel

Multiple-input multiple output systems (MIMO) improve the performance of wireless channels through the use of space-time coding or through spatial multiplexing techniques. As the number of antennae increases, hardware costs naturally increase, as do power requirements. The physical antennas are themselves inexpensive, but the main driving factor is the requirement for multiple parallel RF chains at transmitter, receiver or both. There is consequently interest in developing so-called hybridselection schemes, where the best L out of N antenna signals are chosen at either end. In this way, only the best set of antennae are used, while the remaining antennae are not employed, thus reducing the number of required RF chains, but preferably gaining some channel benefit from their locations as they are switched into and out of the link dynamically. A critical problem faced by such closed-loop MIMO systems in mobile channels is that channel information estimated at a receiver becomes outdated quickly, due to feedback and processing delays. For receive-directed transmit antenna selection schemes, BER is thus sensitive to inaccurate CSI due to feedback delays from receiver to transmitter. In this paper a transmit antenna selection strategy with prediction is investigated for a temporally correlated flat fading Raleigh channel based on minimum error rate criteria, which aims to save transmit power through only transmitting over the best channel -- a suboptimal approach to space-time coding, but one which can be far more power efficient. The minimum error rate method finds particular application in an adaptive coding and modulation network where receive signal strength indication (RSSI) is either not present or not relevant.