Addressing Intersymbol Interference and Nonlinearity in Quadrature Amplitude Modulation Systems using Functional Link Artificial Neural Network Equalizer

This paper proposes Functional Link Artificial Neural Network (FLANN) based equalizers to alleviate the impacts of Intersymbol Interference (ISI) and nonlinearity in Additive White Gaussian Noise (AWGN) corrupted digital communication systems employing 16 and 64 Quadrature Amplitude Modulation (QAM). FLANN offers computational efficiency and does not need hidden layers which makes it an attractive alternative to conventional equalizers as well as equalizers based on Multilayer Perceptron (MLP) neural networks. This paper extends prior published work by considering the effects of nonlinearity and employing a hyperbolic tangent-based multilevel activation function tailored to the specific QAM schemes considered. The channel’s dispersive nature is modelled using a Finite Impulse Response (FIR) filter. Results clearly demonstrate the suitability of FLANN-based equalizers in mitigating channel irregularities.