Sub-Band Cramer-Rao Bounds for Frequency-Selective Spectral Analysis
2005 (English)In: Proceedings of the 39th Asilomar Conference on Signals, Systems, and Computers, 2005, 1439-1442 p.Conference paper (Refereed)
The computational complexity of a parametric spectral analysis method usually increases rapidly for large numbers of measured signal samples. Splitting the considered frequency range into several sub-bands and estimating only the parameters of the signal components in one frequency interval at the time is one possible approach to lower that computational burden. Since the estimation accuracies of such frequency-selective techniques usually differ from those of the corresponding full-band methods it seems unsound to compare the parameter estimates obtained by using a frequency-selective approach to the full-band Cramer-Rao bounds. In this paper we derive sub-band Cramer-Rao bounds as performance bounds for frequency-selective parameter estimation techniques applied to exponentially damped sinusoidal data. We present a numerical study where we compare the sub-band Cramer-Rao bounds to the full-band Cramer-Rao bounds for the estimated frequency, damping, amplitude, and phase parameters.
Place, publisher, year, edition, pages
2005. 1439-1442 p.
IdentifiersURN: urn:nbn:se:uu:diva-76431OAI: oai:DiVA.org:uu-76431DiVA: diva2:104343