Compensation of Nonlinear Distortions in OFDM using FLNN and SOM

H.M.S.B. Senevirathna, K. Yamashita, and H. Lin (Japan)


OFDM, Neural networks, nonlinear distortions, compensation


High power amplifier (HPA) is a main source of introducing nonlinear distortions in an orthogonal frequency division multiplexing (OFDM) system. Methods of improving the power efficiency of a nonlinear amplifier include various amplifier linearization techniques. However amplifier linearization is expensive, power inefficient and HPA dependent. Traditional equalizers not well perform in reduction of bit error rate in presence of such nonlinear distortions in OFDM systems. This paper presents a new method of HPA independent adaptive nonlinear distortion compensator for OFDM signals, which operates at the OFDM receiver end. The compensator is based on a cascade version of a functional link neural network (FLNN) and a self organizing map (SOM), which compensates the HPA nonlinearities with memory effect. Finally we simulate the proposed system in a time dispersive channel with the effect of intersymbol interference (ISI) and with the HPA nonlinearity. Simulation results show the validity of the proposed equalizer.

Important Links:

Go Back