Fast Successive Spectral Estimation of Irregularly Sampled Data

Techniques for estimation of a dense spectrum from irregularly sampled data are typically either very processing intensive or suffer from large amounts of bias in the estimate due to signal leakage. This paper proposes an estimation algorithm that has similarities to the successive interference cancellation algorithms from the communications literature. The algorithm successively estimates larger amplitude frequency components first and then subtracts out those estimates before continuing on to lower amplitudes. The algorithm is able to maintain the low complexity of an FFT-based algorithm while overcoming the poor bias performance typically associated with those algorithms.