Hi,
First thanks for Kent Leung to answer my question.But I think the answer is not what I want.Now I clarify my question again.I have got the decrete sample data and have got the frequency domain waveform in which there are many narrow-band signals.Some signals may have two peaks.So the method Mr.kent suggested may get the wrong bandwidth.I want to measure the bandwidth of each narrow band signal.In my opinion,It is very important to choose proper reference level.I wonder if I have clarified the problem.
Thanks again
