Modulation Recognition of MFSK Signals Based on Multifractal Spectrum

In order to solve the problem that modulation recognition of MFSK signals in alpha-stable distribution noise, a novel algorithm using multifractal spectrum is proposed. Multifractal spectrum characteristics of signals and noise are discussed firstly. Then algorithm extracts the difference between maximum and minimum values of spectrum as classification feature. Finally, algorithm employs threshold decision method to achieve modulation recognition of 2FSK, 4FSK and 8FSK signals. Numerical results show that algorithm has good performance in both alpha-stable distribution noise and Gaussian noise, and it is less affected by characteristic exponent of noise and data points.