Thomson’s Multitaper Method and Applications to Channel Modeling

Abstract

One of the most basic applications of Fourier analysis is power spectrum estimation. Some historical comments on this age-old problem can be found in review articles by Robinson [279] and Benedetto [23], and in Percival and Walden’s book [264]. We refer to Appendix A for basic definitions and properties, and references for stochastic processes. In the problem of estimating a wide-sense stationary random process from data, one has only finitely many samples, often of a single realization of the process, with which to work.