Blind Equalization Using the Support Vector Regression via PDF Error Function
Yang Wang, Ling Yang, Fang Wang, Lu Bai · 2016
In this paper, a new blind equalization is addressed based on support vector regression (SVR) for single-input single-out (SISO) channels, which combines the conventional cost function of the SVR with probability density function (PDF) error function. Based on the iterative re-weighted least square (IRWLS) it solves the equalizer coefficients. Simulation performances show that the proposed equalization method performs better than the traditional algorithms such as constant-modulus algorithm (CMA), PDF algorithm and previous SVR algorithm (SVM-Godard, SVR-Sato) under the intersymbol interference (ISI) standard.