A method of specific emitter verification based on CSDA and SVDD

Specific Emitter Verification (SEV) is an application of feature extraction and classification for determining unknown emitters belong in a certain emitter set from received signal waves. One challenge of this task is the lack of negative samples which makes it difficult to get effective fingerprint and give a good description of the target emitter set. In this paper, a novel Subclass Discriminant Analysis (SDA) algorithm based on C-Means Clustering (CSDA) is developed to get effective feature vectors of the data set and Support Vector Data Description (SVDD) is used to give a description of these fingerprint vectors. Experiment results show the proposed method gets good verification accuracy.