Neural Conjuncted Polynomial's Structure Adaptive Equalizer

Based on the analysis of linear conjuncted polynomial filter and the characteristic of single layer neural network with mild nonlinear and severe nonlinear distortions, novel conjunction's structure equalizer is proposed in this paper, and adaptive algorithm is deduced by the normalized least mean squares (NLMS). Computer simulations show that not only the novel type structure equalizer is simpler in structure and but also can availably remove nonlinear distortions and intersymbol interference (ISI), improve performance of bit error rates (BER) no matter what linear channel or nonlinear channel in digital communication systems.