QPSK error vector magnitude demodulation with RBF neural network in Rayleigh fading channels

Abstract

This work proposes a method to enhance the demodulation of QPSK error vector magnitude (EVM) in a direct conversion receiver (DCR). The proposed method is a radial basis function (RBF) neural network, which is used to learn the characteristics of signal constellation. The hybrid learning method is used to train the RBF network. The hidden layer is trained by the hard k means clustering and the supervised learning is used to train the output layer with given input-output pairs. This study is worked in a Rayleigh fading channel.