Spectral Analysis

Abstract

The estimation of power spectral density (PSD) of a stationary random signal given by a single sample function is known as spectral analysis. The estimation is based on the time average methods assuming that the signal is ergodic. The methods of spectral analysis are also applicable to deterministic as well as nonstationary signals with appropriate modifications. In radar signal processing, the PSD of the echo signal provides the position, intensity and velocity of the moving object by determining the width, height, and location of the peak of the PSD. In passive sonar signal processing, the position of the peak of the PSD gives the direction (bearing angle) of the torpedo. In modeling the propagation channel of turbulent media, the PSD characterizes the decorrelation time, frequency, and distances. In biomedical engineering, the spike- and wavelike form of the PSD indicates the period of epileptic seizure. In speech processing, spectral analysis is used to locate the formants of utterances of particular sound. In electronic warfare, the PSD is used to classify the targets.