Automated Detection and Identification of Cyclostationary Signals

This paper focuses on the detection and identification of cyclostationary signals, which have significant applications across various engineering fields. Accurate detection and characterization of these signals is crucial for many practical implementations. The core contribution is an automated methodology implemented through the autofam function, which performs feature extraction and analysis on input signals. The function signature autofam(x, fs, df, dalpha) requires four key parameters: the input signal x, sampling frequency fs, frequency resolution df, and cycle frequency resolution dalpha. These parameters control the algorithm's spectral resolution and modulation characteristics analysis. The implementation employs cyclostationary signal processing techniques to compute the spectral correlation density function Sx, while simultaneously estimating the modulation index alphao and center frequency fo. These outputs provide comprehensive characterization of the signal's periodic properties and modulation characteristics. The algorithm operates by analyzing second-order periodicity in signals through spectral correlation analysis, making it particularly effective for detecting hidden periodicities in modulated signals. The function implementation includes optimization for computational efficiency while maintaining accurate feature extraction. This automated approach enables efficient analysis of cyclostationary properties without manual parameter tuning, making it suitable for real-time applications and batch processing scenarios. The methodology holds significant promise for applications in communications, radar, mechanical fault detection, and other domains where periodic signal characteristics need to be identified and analyzed.