Simplified ANN for 256 QAM Symbol Equalization Over OFDM Rayleigh Channel

Increasing the specter efficiency has been an object for many studies. In this paper, we investigate the higher modulation 256 QAM using Artificial Neural Networks (ANN) as an equalization model. Multilayer perceptron (MLP) and Radial Basis Function (RBF) are considered as non-linear equalizer based on back-propagation and Euclidian norm respectively. They are designed in a simplified architecture and employing some performing strategies for a better learning and an increased processing speed. ANNs are presented and applied with Orthogonal Frequency Division Multiplexing (OFDM) over Rayleigh fading channel in order to optimize the modulation scheme's processing and performances despite its sensitivity to noise. The models will be compared to the theoretical BER simulation in terms of BER, and also in terms of MSE to show performance and efficiency; by that, this work will show the supremacy of MLP in decision making with 256 QAM.