SVM Based Classification and Fusion Algorithm of Steady-state Signal Features of Radiation Source

The identification of subtle characteristics of steady-state signals from radiation sources is a hot and difficult research topic. In this paper, fingerprint features of different power amplifiers are extracted based on bispectral analysis, fractal dimension and HHT. Aiming at the problem of poor identification effect of single feature, SVM is adopted to classify and fuse different features. Theoretical analysis and experimental verification show that the steady state signal feature classification fusion algorithm is superior to the single feature classification algorithm in identification performance.