Exploring New Features for a Wavelet Neural Digital Modulation Recognition System-eng

Abstract

Modulation recognition has been an important problem in both commercial and military wireless communication. Modulation recognition can be divided into two categories: identification between categories and identification in category. In this work a system is proposed for identification between categories of different digital modulated signals using a combination of discrete wavelet transform (DWT) and the linear predictive coding (LPC) with the probabilistic neural network (PNN) as a classification tool. It was found that the proposed system out performed any of the existing systems by using six DWT decomposition levels and 20 LPC coefficients. The symlet 20 wavelet filter proved to be the best candidate. The results showed that a 100% recognition can be achieved at a signal to noise ratio (SNR) of 2db for the digitally modulated signal.