Fractionally spaced channel equalization using non-linear PCA

In this article we introduce a new blind fractionally spaced channel equalizer based on higher order statistics using non linear principal component analysis (NLPCA). The input symbol sequence are taken as independent identically distributed with non zero fourth order cumulants. We formulate the fractionally spaced equalizer into blind source separation (BSS) problem and solve using NLPCA of BSS and compare the results with the method of cumulants. Simulation results show that the performance of fractionally spaced channel equalizer is better in non linear principal component analysis environment.