Comparison of higher-order and cyclic approaches for estimating random amplitude modulated harmonics

To estimate harmonics observed in random multiplicative and additive noise, algorithms relying upon second- or higher-order statistics have been derived after stationarizing the available cyclostationary and nonGaussian time series. Recently developed cyclic approaches exploit cyclostationarity and rely upon the cyclic mean, cyclic covariance, or, higher-order cyclic cumulants in order to: (i) detect and estimate frequencies of mono- and multicomponent random amplitude modulated harmonics, and (ii) characterize the multiplicative processes. Comparison based on estimator variances and simulations illustrate that cyclic approaches surmount existing methods in SNR gains and resolution and obviate restrictions on the bandwidth and distributions of the additive and multiplicative noise.