A DSP based frequency domain approach for non-stationary signals using S-Transform

Abstract

Non-stationary time series signals such as the power system signals, speech signals, and genomic signal are gaining more importance as they pertains to information unspecified in time-domain analysis. In-depth study of such signals require a domain (Frequency domain), which could aid in identifying this unspecified information. Fourier transform (FT) is an eminent technique to analyze the frequency content of a stationary signal. For a non-stationary signal, frequency content changes with time, where FT fails to arrest the time sprouting effects of the frequency. Time-frequency representation (TFR) is of importance to analyze such signals. Short-Time Fourier Transform (STFT) and Continuous Wavelet Transform (CWT), are the techniques pertaining TFR of a signal, can partly alleviate the problem. S-Transform is a phase corrected form of CWT. S-transform is a powerful tool for detection and localization of time frequency components of signals. Superiority of S-Transform is shown over the STFT and CWT when applied to Non-stationary signals.