Power spectral density estimation simulation using the maximum entropy method

摘要

This paper presents the results of computer simulations using the maximum entropy method (MEM) for power spectral density estimation. The correlogram method was also compared. The Burg algorithm was shown to be superior for resolving sharp spectral peaks, especially for short sequences. Windowing of the error sequence for the Burg algorithm was shown, in some cases, to significantly reduce the frequency bias of sinusoidal peaks.