Selection of data windows for digital signal processing

In this paper study on date windows has been presented which enables to select a window optimally for a desired application in digital signal processing. Various window parameters for a number of windows have been compared. The effect of quantization of window-samples on these parameters have been determined which show that the window-sample word-length of 8 bits is adequate to closely approximate the unquantized window. The windowing in frequency domain has been considered end compared with the time domain implementation for spectral estimation (using FFT) applications. It is seen that the former, in general, is simpler for implementation and provides a better computational accuracy.