Optimum Sampling Times for Spectral Estimation.
Abstract
The problem of optimum sampling strategies for spectral estimation of Fourier-type signals in the case of finite discrete-time observations was investigated. In particular it was shown that minimum variance unbiased estimates of amplitudes of sine and cosine terms of Fourier signals embedded in additive zero-mean white noise can be determined by sampling at the generalized Chebyshev times. The solution obtained, by putting the problem in an optimum linear regression framework, is that the generalized Chebyshev times are the zeros of the derivative of the highest frequency cosine wave. If the number of samples exceeds the intrinsic dimensionality, repeated independent sampling at those points not only provides the best approximation to the Fourier signal in a minimum variance sense but also the linear minimum variance unbiased estimate of the coefficients.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 30, 1982
- Accession Number
- ADA114871
Entities
People
- Lonnie C. Ludeman
Organizations
- New Mexico State University