Window selection for the multi-window periodogram spectrum estimation and spectrogram time-frequency analysis

Abstract

The concept of multi-window or multi-taper spectral estimation was discussed in [1]-[5]. This is performed by averaging the spectrum of the signal modulated with any set of orthogonal windows [1], such as minimum bias windows, sinusoidal windows and the sequence of discrete prolate spheroids or the Slepian sequence [2]. The Slepian sequence is the most common orthogonal set of windows used in the spectral analysis. It was first introduced in [3] where it is applied to stationary spectral analysis. It is shown in [4] that this technique have better performance in terms of variance, leakage and resolution as compared to the weighted overlapped segment averaging method (Welch periodogram).