A new feature vector using local surrounding-line integral bispectra for identifying radio transmitters
Shuhua Xu, Benxiong Huang, Zhengguang Xu, Yuchun Huang · 2007
A novel method for identifying radio transmitters with the same model and manufacturing lot is proposed in this paper. The local surrounding-line integral bispectra are selected through the Fisher’s class-separability discriminant measure as the main feature parameters, and they are interfused with parameters significant for classification of the received signal to form a new identification feature vector. A radial basis function(RBF) neural network is implemented to realize classification and identification for the individual transmitter utilizing the new feature vector. The selected features are evaluated using sample data of ten FM stations with the same model and manufacturing lot. It is shown that they are highly discriminative even in low SNR.