Drone Identification Based on Normalized Cyclic Prefix Correlation Spectrum

• DOI: 10.1109/tccn.2024.3375514
• Publication Date: 2024-03-19T18:35:10Z
• Authors: Hanshuo Zhang; Tao Li; Nan Su; Dishan Wei; Yongzhao Li; Zhijin Wen
• Keywords: Cyclic prefix; Drone; Robustness

Utilizing deep learning (DL) to identify drones through radio signals has been proven be a promising approach. However, two significant challenges remain solved. The first is how effectively at the low signal-to-noise ratio (SNR) regime, and second stably among numerous unknown interferences. In theory, sufficient data can alleviate above problems, but costs of signal acquisition labeling are usually unacceptable. this work, we aim improve robustness feature representation by introducing stable prior knowledge drone problems. Since commonly adopt orthogonal frequency division multiplexing (OFDM) modulation with particular cyclic prefix (CP) structures for video transmission, propose identification algorithm using convolutional neural network (CNN) normalized CP correlation spectrum (NCPCS). NCPCS strongly correlated invariant parameters, i.e., OFDM symbol duration duration, mutually exclusive other signals. Thus, natively improves system interference Besides, keep characteristic clear SNR calculate improved accumulating multiple consecutive symbols. increase in length sharpens peaks SNRs. Finally, suitable CNN augmentation (DA) method proposed precisely generically extract these characteristics from drones. universal software peripheral (USRP) X310 utilized collect five construct experimental dataset. We test under different conditions: Gaussian noise condition co-frequency condition. Experimental results show that outperforms waveform-based, spectrum-based spectrogram-based algorithms conditons. remains good