The Multimode Blind Equalization Algorithm Based on Gaussian Process Regression

A new multimode blind equalizer based on Gaussian process regression(GPR) by incorporating multimode algorithm(MMA)-like error function into the conventional GPR framework is proposed. The MMA algorithm will be introduced to the phase information, error function can be divided into two parts of the real part and imaginary part, so as to solve the problem of the phase rotation. The GPR framework formulates the posterior density function for weights using Bayes' rule under the assumption of Gaussian prior for weights. The proposed blind GPR equalizer is based on linear-in-weights regression model. The simulation results show that the proposed blind GPR equalizer not only without cumbersome cross-validation procedures compared to the state-of-the-art blind support vector machine (SVM) equalizer, but also shows the better performances in terms of intersymbol interference and the constellation diagram than the CMA, MMA, SVR equalizers in linear and complex channels.