A Novel Analog Modulation Classification: Discrete Wavelet Transform-Extreme Learning Machine (DWT-ELM)

Abstract

The aim of this study is to propose a method using discrete wavelet transform and extreme learning machine (DWT-ELM) in classification of communication signals. Six types of analog modulated signals as “AM”, “DSB”, “USB”, “LSB”, “FM” and “PM” are used for classification and analog modulated signal dataset consists of 1920 signals. These signals are also added white noise. Feature extraction is performed using different DWT filters. The feature vector obtained from DWT is used in classification. ELM is preferred due to its advantages over conventional back-propagation based classification. The feature vector is fed by the inputs of the ELM. The performance of the proposed method is evaluated for different types of DWT filters. In addition, compared results with similar study are presented to be able to determine the success of the proposed method.