Título: A Simple Recurrent Neural Network Equalizer Structure
Autores: Coelho, Pedro Henrique Gouvêa
Resumo: This paper describes a one-processing-neuron recurrent neural network for application in channel equalization using variations on the RTRL (Real Time Recurrent Learning ) algorithm for training the neural network. The structure is very simple and its computational demand is very low due to the use of only one processing neuron in its architecture. Simulation results are presented including several cases involving BPSK, 8PSK and 16PSK modulation schemes in additive Gaussian noise.
Código DOI: 10.21528/CBRN2001-106
Artigo em pdf: 5cbrn_106.pdf
Arquivo BibTex: 5cbrn_106.bib