Radiation Source Identification based on Box Dimension of PF and SIB Fusion
Shuang Ma, Yanhua Jin, Xin Hua Zhou · 2019 IEEE International Conference on Signal, Information and Data Processing (ICSIDP) · 2019
To solve the problem that single characteristic of radio transmitters in low-end receivers is not ideal for the individual identification of radiation sources, the algorithm that box dimension of Product Function (PF) decomposed by Local Mean Decomposition(LMD) fuses Rectangular Integral Bispectrum(SIB) is proposed. Firstly, the acquired time domain signal is decomposed by LMD to obtain a set of PFs, and the box dimension of the PF is used as the radiation source feature. Then, the SIB is estimated by the signal high-order spectrum, and the SIB after dimensionality reduction is taken as the feature. By Canonical Correlation Analysis (CCA), the two features are fused as the fusion feature, which is identified by Support Vector Machine (SVM). Finally, this algorithm is validated in experiments where USRP-2920 is regarded as the signal receiver. In the classification experiment of 6 walkie-talkies as signal source devices, the recognition rate of the fusion feature is 13%~28% higher than the single feature. This paper applies the LMD algorithm to the field of RF radiation source identification for the first time.