A phase-locking technique for estimating the ensemble average of time-series data
Barry S. Perlman, V H Auerbach · IEEE Transactions on Acoustics Speech and Signal Processing · 1977
A digital method employing Fourier analysis is presented, which in the absence of any external synchronizing signals can, with certain restrictions, 1) reduce random noise from repetitive frames of sampled input data; and 2) average waveforms which may differ in shape and which may occur in the presence or absence of random noise. The restrictions refer to the ability to select a window width which will each time encompass all of the aperiodic waveforms or one period of a periodic waveform. There is no restriction as to where in the window any waveform may appear. The new method is called the phase-locked time (PLT) average and should prove useful in a variety of situations.