Neural networks as nonlinear compensator models for digital communication systems

An important element of digital communication systems is a power amplifier (PA). PA is the source of nonlinear distortions at its high energy performance. PA nonlinear characteristics produce the expansion of signal spectrums that leads to interchannel interferences. One of effective ways of nonlinear distortion cancellation is nonlinear compensator synthesis. There are two types of compensator models, namely polynomials and neural networks. The functional link artificial neural network (FLANN), polynomial perceptron network (PPN) and radially pruned Volterra model are supposed as compensator models. These models and compared according to provided accuracy and computational complexity.