An adaptive radius blind equalization algorithm based on pdf fitting

In this paper, we investigate blind equalization techniques based on information theoretic criteria. They involve estimating the probability density function (pdf) of transmitted data. Our work is based on previous studies where the Parzen window method has been used to estimate the pdf at the equalizer output. The equalizer is obtained by minimizing the distance between this equalized pdf and some target distribution. With a view to reduce algorithm complexity, we propose a reduced constellation implementation of the adaptive equalizer. We show complexity and performance gain against similar approaches in the literature.