Tuning the windowed Fourier transform of robotic signals in the perspective of information theory

In many applications we are interested in the frequency content of a signal at a given period of time. The windowed Fourier transform is one of the most widely used time-frequency representations. To use this technique several parameters must be defined according to the signal analyzed. A new method based on the mutual information between frequency and time is proposed. The tests developed with robotic signal show the capability of the presented method.