An improved algorithm for direct time-domain averaging

Time-domain averaging (TDA) is a common method to extract a periodic compon ent of interest from a noisy compound signal. With the period of the intere sting component determined, we often consecutively cut out some segments in length of the period from the compound signal and directly average them. I n such direct averaging with digital computers, the period actually used po ssesses some error, called period cutting error, which is less than half a sampling interval. This paper finds that there is a large bad influence of the period cutting error on the direct TDA results. Furthermore, the paper introduces an improved algorithm for the direct TDA. The improved algorithm basically solves the problem caused by the period cutting error. With some tests and analyses, it was shown that the improved algorithm is much bette r than the original one. (C) 2000 Academic Press.