Noise Recognition in Communication Channels by Artificial Neural Networks

Abstract

In this paper, the processes about synthesis of artificial neural networks for noise recognition in communication channels were considered. Gaussian White Noise (GWN) and Periodic Random Noise (PRN) as well as digital signals with the presence of these noises are objects of simulation investigation. Accuracy, Mean Squared Error (MSE), regression and ROC curves and specific indicators for classification quality are analyzed. A different number of neurons in the hidden layers and type of output transfer function are evaluated. The accuracies more than 90.00 % were observed about the test signals.

Keywords- noise recognition, GWN, PRN, IFN, artificial neural networks.