A Study on Modulation Signal Recognition Using Differential Constellation Trace Figure

This paper presents a novel approach to modulation signal recognition method for wireless devices. Specifically, we propose a differential constellation trace figure (DCTF) scheme for extracting the features from I/Q samples, which are subsequently utilized by a neural network algorithm to enable device identification. To comprehensively evaluate the efficacy of the DCTF scheme, we simulated and tested 12 distinct modulation signals in various environments characterized by carrier frequency offsets and signal-to-noise ratios (SNRs) spanning from 0 to 20dB. The experimental results reveal that the proposed DCTF scheme achieves a recognition accuracy exceeding 99% in scenarios where the SNR surpasses 12dB, even in the presence of high carrier frequency offset, thereby outperforming state-of-the-art recognition schemes such as IQ-CNN, IQ-LSTM, and CLDNN.