A wavelet-based method for classification of binary digitally modulated signals

  • Authors:

  • Ka Mun Ho;Canute Vaz;David G. Daut

  • Affiliations:

  • Department of Electrical and Computer Engineering, Rutgers, The State University of New Jersey, Piscataway, NJ;Department of Electrical and Computer Engineering, Rutgers, The State University of New Jersey, Piscataway, NJ;Department of Electrical and Computer Engineering, Rutgers, The State University of New Jersey, Piscataway, NJ

  • Venue:
  • SARNOFF'09 Proceedings of the 32nd international conference on Sarnoff symposium
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this study, a wavelet transform-based technique is used in an Automatic Modulation Recognition (AMR) process to classify different types of digitally modulated binary signals. The communications signals considered are Binary Amplitude Shift Keyed (BASK), Binary Frequency Shift Keyed (BFSK), and Binary Phase Shift Keyed (BPSK) signals, which are transmitted over an Additive White Gaussian Noise (AWGN) channel having a Signal-to-Noise Ratio (SNR) in the range from -5 dB to 10 dB. The distinguishing features of these three modulation schemes arise due to variations of amplitude, frequency and phase of a carrier signal. The different types of binary communications signals are analyzed using the Continuous Wavelet Transform (CWT). The unique features of each modulation type are extracted from the specific wavelet-domain representation of the respective signals. The features are stored as templates within the receiver and used for the purpose of classifying the signal according to modulation type. The wavelet used for template construction and the decomposition of received signals is the Reverse Biorthogonal Spline 1.3 (rbio1.3) wavelet. It has been determined via extensive computer simulations that the rate of correct classification for BASK signals is 100% and for BPSK signals is 99.7% over the range of SNR values considered. The rates of correct classification for BFSK signals are 99.6%, 98.7%, 94.0%, and 54.0% for SNR = 10 dB, 5 dB, 0 dB, and -5 dB, respectively. The AMR process presented in this study generally produces higher rates of correct classification than other AMR techniques available in the literature. This observation is especially significant when considering the cases of BASK and BPSK for systems operating at an SNR value of -5 dB.