Optimal Short-Time Fourier Transform's Parameters and Gaussian window for Enhancing Signal Separation

The observation of dynamic systems is essential in many fields in signal separation. Many algorithms are used to estimate this dynamic system, consequently, estimating and recognizing its dynamic parts. Short-Time Fourier transform (STFT) is one of these algorithms. This paper presents optimized STFT parameters to extract the Doppler properties of that system. The gradient descent optimization approach, Adaptive Movement Estimation (ADAM), is used. It is simple and fasts than the other to achieve the best solution. A Gaussian window has been selected to be used, while, the improvement reached 7-11% against the unoptimized process, and the processing speed was also affected by 35%.