A multitime-frequency approach for detection and classification of noisy frequency modulations

In this communication, time varying higher order spectra and specifically multitime-frequency representations have been used for detection and classification purposes. A new detector is presented for frequency modulation disrupted by multiplicative and additive noise. A statistical study is performed and corresponding simulations are presented. An extension to multiple hypothesis testing is also presented to classify neighboring frequency modulations in a context of multiplicative and additive noise. Some simulations illustrate the performance of our approach compared to the similar second order approach.