Individual Recognition for Satellite Communication Based on Feature Fusion Method

Abstract

A novel algorithm based on signal's steady state feature fusion method is proposed in this work to deal with the individual recognition for satellite communication, which tries to combine the extracted wavelet coefficients, bispectrum features and fractal dimension of sampled data with canonical correlation analysis (CCA). The popular kernel principal components analysis (KPCA) method is thereafter introduced to reduce the feature dimension and several familiar classifiers are designed to verify the results. It has proved that the fused features for specific individual satellite communication terminals holds evident separability that are available for traditional classification algorithm. The experiments have demonstrated the excellent performance of proposed method.