ABSTRACT
Frequency-domain methods, such as Fourier transform, are useful in science and engineering. This chapter covers the characterization of random processes in frequency domain. It discusses the concept of power spectral density, properties of power spectral density, and white noise. The chapter explains power spectrum estimation, and cross-power spectrum and its properties. Any random process with a power spectrum that has a fiat top or sharp comer is not really physically realizable. An important issue in practice is how to estimate the power spectrum of a stationary random process, given a finite-length data record of the process. There are two classes of methods for power spectrum estimation: parametric and nonpara-metric. A parametric method estimate the power spectrum of a random process based on an assumed model for the process and is thus also known as a model-based method.
