Vision Electrónica, ISSN 1909-9746, 2248-4728, Volume 11, 2017, pp. 1-20, Issue published: March 1, 2018

https://revistas.udistrital.edu.co/index.php/visele/article/view/14624

Given a sampled signal, in general, is not possible to compute its period, but just an approximation. We propose an algorithm to approximate the period, based on the Discrete Fourier Transform. If that transformation uses data length for multiples of the true period, some of its harmonics have null value. Thus, the best candidate to be a multiple of the period minimizes the value of those harmonics. The validation for noiseless data shows an upper bound in the error equal to a quarter of the time between two consecutive samples, whereas the result for noisy data demonstrates robustness. As application, the algorithm estimates the period of physiological signals, and tracks the frequency of the power grid in real time, which evidence its versatility.