Spectral Analysis of Unequally Spaced Data Samples,
Abstract
Most of the theory and practice of digital spectral analysis is based on use of the discrete Fourier transform, which by definition requires a set of data values measured at equally spaced times. In the paper the authors consider the analysis of data that are only approximately on an equally spaced grid and that in addition suffers from the problem of missing observations. In this case, the discrete Fourier transform is not directly applicable, and the authors consider several alternative approaches. The authors then consider in detail the measurement of the frequency of a single sinusiodal function in the presence of noise, using an approach based on minimizing the residual squared error after fitting. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1972
- Accession Number
- AD0743497
Entities
People
- A. G. Larson
- R. C. Singleton
Organizations
- SRI International