A nonlinear equalizer based on estimation of RBF´s centers

An RBF equalizer has some vectors called centers and these vectors are generally calculated by a probability density function of received signal. If a characteristic of the channel changes, we can´t optimally estimate a transmitted signal by this equalizer. In this paper, We assume that a transmitted signal is two-level value. We firstly design two RBF equalizer, namely the optimal RBF equalizer when the characteristics of channel are steady, and the optimal RBF equalizer after the characteristics of channel change. We show that the parameters are the centers which clarify the distinction of the property of the two RBFs above. Secondly, with this result, we propose the RBF equalizer where centers are estimated with LMS algorithm and of which another parameter is fixed. Then we verify that the proposed equalizer is available by computer simulation