Fourier Meets Gardner : Robust Blind Waveform Characterization

Waveform Characterization is crucial for various spectrum sensing applications such as anomaly detection and measuring spectrum utilization. It consists of detecting the waveform type (single carrier or spread spectrum), modulation form (QAM, PSK, FSK, GMSK, GFSK etc ̇) and corresponding parameters such as symbol rate and chip rate. In this paper, we propose a blind characterization algorithm suited for these applications using second-order cyclostationary and fourier domain features of signals. To test the proposed method’s robustness, a comprehensive evaluation is conducted using both simulated and over-the-air (OTA) experiments with appropriate signal detection pre-processing steps. An overall modulation classification accuracy of 86.25% is attained for OTA testing with a modulation set consisting of QAM, PSK, FSK, GFSK, MSK, GMSK, DSSS and OOK.