An algorithm for separating multilevel random telegraph signal from 1f noise

In this work, we propose a robust algorithm for the separation of two- and multilevel dominant random telegraph signals (RTSs) from 1f noise in the time domain. The method does not associate each RTS level to a fixed range of the signal values, as assumed by other methods, but it is based on a efficient recognition of the jumps between the different RTS levels. The proposed algorithm can extract the 1f component even in the presence of several dominant RTSs with different corner frequencies. The procedure has been validated by using a two-level and a four-level synthesized signals. © 2008 American Institute of Physics.