Signal analysis and classification of digital communication signal using higher order time-frequency analysis techniques

Spectrum monitoring is important for spectrum management and intelligence gathering especially in non-cooperative environment. There are various methods proposed on modulation classification. However, these methods give low classification accuracy at low signal-to-noise ratio (SNR). In nature, signals are exposed to both noise interference and multipath fading. These impairments make the signal analysis and classification even more difficult. In this paper, the adaptive smooth-windowed Wigner Ville bispectrum (SWWVB) which combines the advantage of time-frequency analysis and higher order statistics (HOS) is proposed. The modulation type and parameters are estimated from the time frequency representation. A rule based classifier is then used to classify the typical digital modulation signal such as ASK, FSK and M-ary FSK. The adaptive SWWVB is able to maintain consistent classification accuracy in both noise and fading condition.