High resolution autoregressive spectrum analysis using noise power cancellation

Autoregressive spectrum analysis, alternately known as maximum entropy spectrum analysis, has received much attention as a higher resolution alternative to conventional spectra computed using the modulus of the FFT. This paper examines the relationship of the autoregressive spectrum analysis technique to the Prony and Pisarenko spectral decomposition techniques. The Prony method yields a new signal power estimation procedure for AR spectrum analysis and the Pisarenko method yields a noise power estimation technique for use with AR noise power cancellation spectrum analysis.