A method for the calculation of spectrum samples of discrete, aperiodic and finite signals based on the DTFT is proposed. This method is based on a flexible discretization of the frequency variable that could produce equidistant, sparse or unique spectrum samples. It is implemented in a GPU platform as a Matrix-Vector product, being able to be applied on modern HPC systems. As a result, a general use tool is developed for the frequency analysis that achieves execution times in a linear relation with the length of the vector to be processed and the number of samples required. Finally, it is shown that the required execution time for the computation of equally spaced spectrum samples is competitive to the achievements of other tools for frequency analysis based on sequential execution.

Leopoldo Noel Gaxiola-Sánchez

Papers

Spectrum Sample Calculation of Discrete, Aperiodic and Finite Signals Using the Discrete Time Fourier Transform (DTFT)