Automatic modulation recognition for MFSK using modified covariance method

Abstract

This paper presents modulation classification method capable of classifying MFSK digital signals without a priori information using modified covariance method. This method using for calculation features for FSK modulation should have a good properties of sensitive with FSK modulation index and insensitive with signal to noise ratio SNR variation. The numerical simulations and investigation of the performance by the support vectors machine one against all (SVM-OAA) as a classifier for classifying 6 digitally modulated signals which gives probability of correction classification up to 85.85 at SNR=-15dB.