Abstract
Chi and Chen proposed a fast kurtosis maximization algorithm (FKMA) with computationally efficient for blind beamforming, blind maximum ratio combining, and blind source separation or independent component analysis. The FKMA assumes that each source signal is non-Gaussian independent identically distributed process and statistically independent with each other. In uplink orthogonal frequency division multiplexing (OFDM) systems with antenna arrays, we establish a discrete-time multi-input multi-output signal model for pre-FFT and post-FFT beamforming receiver, respectively. For the former, due to the IFFT process at the transmitter, the received time-domain source signals are approximate Gaussian distributed statistically such that the performance of the FKMA degrades seriously. For the latter, due to the FFT process at each antenna of the receiver, the equivalent frequency-domain source signals are non-Gaussian independent non-identically distributed and mutually dependent statistically in multipath. In general, we utilize ensemble average to estimate the source statistic for analyses of an algorithm such as FKMA. However, in the post-FFT beamforming receiver, we observe that the equivalent frequency-domain source signals are mutually orthogonal using the calculation of time average. Therefore, we analyse the FKMA by time average and apply the FKMA to the post-FFT beamforming receiver in multipath; and further, we propose a blind beamforming algorithm (BBA). Finally, some simulation results are presented to support the efficacy of the proposed BBA.