An Lfm Radar Signal Source Identification Method With RFF Drift Robustness

Abstract

Translate Article

Linear frequency modulation(LFM) signals are widely used in radar technology. Accurate identification of LFM signal sources is of great significance. Radio frequency fingerprinting(RFF) is a non-cryptographic authentication method based on hardware physical differences, which is unique and therefore widely used for signal source identification. However, in practical applications, the RFF of radar signal sources generates heat after prolonged operation. This causes a certain degree of drift in the RFF within and between pulses, thereby affecting recognition accuracy. To address this issue, this paper proposes an LFM radar signal source identification method with RFF Drift Robustness(RDR). RDR first uses a frame-based coherent accumulation method to separately calculate the RFF of the front and rear halves of the pulse, then introduces a Class Principal Component Analysis layer(PCALayer) to mitigate the impact of RFF drift within the pulse. Subsequently, an Long Short-Term Memory(LSTM) Network is used to capture the temporal dependency of RFF between pulses, thereby enhancing the model's robustness to RFF drift between pulses. Experimental results demonstrate that the proposed method exhibits excellent robustness to RFF drift in real-world scenarios.