Equal observation interval comparison of maximum entropy and weighted overlapped segment averaging spectrum estimation techniques

This correspondence presents a brief summary of an experimental comparison between conventional spectral estimation methods and a maximum entropy spectrum analysis (MESA) algorithm. These results should be of some interest to workers in acoustical signal processing, especially sonar and surveillance, due to several factors in the form of the experimentation. Specifically, these factors are the range of signal-to-noise ratio (SNR) studied, comparisons based on equal length observation intervals, the use of weighted overlapped segment averaging (WOSA), and the use of ensemble averaging after maximum entropy analysis. Results are presented for resolution and peak signal response and tend to indicate that the MESA method shows promise of achieving the detection performance of long observation interval discrete Fourier transform analysis at a reduced observation time for a useful SNR range.