Maximum Entropy Spectral Analysis: Theory and Application

Abstract

In this paper, I present the rationale for and the theory behind the maximum entropy method of spectral estimation for stochastic processes. The paper is divided into three sections I Introduction, Theory, and Computer Application. In the introduction, a brief review of the two widely used alternate means of spectral estimation is presented. The inherent drawbacks associated with the autocorrelation method and the periodogram method are discussed. The rationale is then presented for the maximum entropy method. The theory behind the maximum ·entropy procedure is developed in the theory section along with the algorithms that will be used in the computer implementation. This section may be.omitted by those whose interests are confined to the application of this method. The section on the computer application of the maximum entropy procedure contains a brief description of each of the computer programs listed in Appendix B along with instructions for their use.