A New Classification Method of FSK Signals Using  Multiband Energy Operators

n this paper， a new classification algorithm of MFSK signal is analyzed. Aiming at the problem of feature extraction， a method based on multiband energy operators is given. This method first filters the signal through a bank of passband filters to reduce the effect of noise， and then applies the energy operators to the largest filter output response to extract the classification character. Aiming at the disadvantage of FCM algorithm which is sensitive to initial values and easily falls into local optimum solution， a novel cluster algorithm based on the kernel method fuzzy kernel C-means clustering algorithm (FKCM) is used; It uses the Mercer kernel mapping the feature vector impliedly to the high dimensional feature space and makes the feature in the input space undistinguishable become discriminable， and improves the clustering performance. Through computer simulation it can be seen that the feature extraction method based on multiband energy operators can restrain noise effectively and FKCM clusters well， so the classification accuracy is high.