A compressive sampling-based method for classification and parameter estimation of FSK signals

The paper presents a new method for the classification and parameter estimation of frequency-shift-keying (FSK) signals, which exploits compressive sampling, and thus, allows to reduce the sampling rate beyond the limit of the Shannon theorem. The method identifies the modulation scheme among the set consisting of 2-FSK, 4-FSK, and 8-FSK, and determines the tone spacing of the modulation. The proposed method has been implemented in GNU/Octave and evaluated both by simulations and experiments on emulated FSK signals, in the presence of additive white Gaussian noise.