Circulant singular spectrum analysis: A new automated procedure for signal extraction

Sometimes, it is of interest to single out the fluctuations associated a given frequency. We propose new variant SSA, Circulant SSA (CiSSA), that allows extract signal any frequency specified beforehand. This novelty when compared with other procedures need identify ex-post frequencies extracted signals. prove CiSSA asymptotically equivalent these alternative although advantage avoiding subsequent identification. check its good performance and compare methods through several simulations for linear nonlinear time series. also validity in nonstationary case. apply two different fields show how works real data find behaves successfully both applications. Finally, we state art techniques used signals amplitude varying time.