Radio Frequency Fingerprinting based on Differential Reconstructed Phase Space

Radio Frequency Fingerprinting (RFF) is the technology to extract the hardware-specific information of the transmitter system by measuring the characteristics of the received signal, thus realizing the identification of the emitter. However, in the receiving process of a receiver, random perturbations including initial phase, frequency offset and amplitude gain, will be attached to the signal, thus affecting the performance of RFF. To solve this problem, we propose a differential reconstructed phase space (D-RPS) algorithm based on nonlinear dynamics and extract nonlinear features based on D-RPS to improve the robustness of the original methods to these random perturbations. Experimental results on simulated and real-world data show that the D-RPS-based feature methods are significantly more robust and effective than the original features.