The price tag for a fast Fourier transform on any sample size

✍ Scribed by James B. Cheek

• Publisher: Elsevier Science
• Year: 1991
• Tongue: English
• Weight: 598 KB
• Volume: 5
• Category: Article
• ISSN: 0888-3270
• DOI: 10.1016/0888-3270(91)90002-m

✦ Synopsis

A computer program that makes a time-to-frequency domain transformation on a signal sampled with any number of equally spaced points has been available for over 10 years. That program, Glassman's discrete Fourier transform (DFT), is not generally used. Perhaps this low use is due to the wide and unpredictable variations in the processing time caused by small changes in the sample size. This paper examines that program and shows how the sample size controls the processing time. The procedure described in this paper establishes the relative computation cost of any sample size and shows that small adjustments to the sample size will provide the DFT capability at an affordable cost. So, this paper is largely about the workings of a computer-based signal processing tool and how to use it effectively.