Blind MIMO-OFDM Channel Estimation Based on ICA and KRLS Algorithm
Haiyuan Liu, Jiancheng Sun · 2009
In this paper, a new method on MIMO-OFDM blind channel estimation based on independent component analysis (ICA) and kernel recursive least squares (KRLS) is proposed. The MIMO-OFDM channel estimation frequency domain expression formula has the similar structure with the blind source separation, so the natural gradient ICA is used to calculate the weight frequency matrix, KRLS is been utilized to compute the others. The method has good ability on line learning, simulation shows that under the similar condition, the performance has about 0.6~0.8 dB higher comparing to subspace and virtual subcarrier method.