Cycle-shift stop-go adaptive filter and its application in the separation of frequency domain overlapped periodic signals

A cycle-shift stop-go adaptive filter (CSSGAF) is proposed to detect the periods and to separate mixed periodic signals automatically. No prerequisite of the signals is provided and little a priori knowledge is required. A method is provided to separate periodic signals which are overlapped both in time and in frequency domain. The convergence of CSSGAF is discussed theoretically in both stop and go stages, and the requirements for the selection of filter parameters are given. A rough scanning and channel autoshut algorithm is designed to reduce the amount of calculation in the searching stage. The function of CSSGAF is tested by using it to separate various kinds of mixed-signals. Experiments show that only part of the sequence of the mixed periodic signal is required, and that the initial phases and relative values of the signals do not affect the separating ability of CSSGAF