Analog Antenna Combining for Maximum Capacity Under OFDM Transmissions

In this paper, we study beamforming schemes for a novel MIMO transceiver, which performs adaptive signal combining in the radio-frequency domain. Assuming perfect channel knowledge at both the transmit and receive sides, we consider the problem of selecting the transmit and receive RF beamformers that maximize the capacity (MaxCAP criterion) of the system under orthogonal frequency division multiplexing (OFDM) transmissions. This problem is non-convex and has no closed-form solution, therefore the maximum capacity beamformers are found using a gradient search algorithm. Furthermore, it is shown in the paper that, for low signal-to-noise ratios (SNR), the MaxCAP criterion is equivalent to maximizing the received SNR (MaxSNR criterion). However, for moderate and high SNRs, the maximum capacity beamformers sacrifice part of the received SNR in order to improve the worst subcarriers and, in this way, they increase the overall capacity of the multicarrier channel. Finally, by means of numerical examples we show that the MaxCAP criterion significantly outperforms the MaxSNR criterion in terms of bit error rate and outage probability.