Tools For Multivariable Spectral and Coherence Analysis
Abstract
The aim of this research has been to develop quantitative tools for high resolution signal analysis and system modeling. These consist of theory and software. An important contribution has been to introduce a family of distance metrics to be used for comparing power spectral distributions in multivariable time-series. These metrics allow quantitative assessment of uncertainty and of resolution in spectral analysis. They also provide a geometric substrate for new techniques in filtering and estimation. More specifically, the concepts of geodesics and that of geometric means, have provided models for time-varying spectra and for ways to aggregate information from disparate sources, respectively. A description of software developed is maintained at the PI's website (http://www.ece.umn.edu/~georgiou/files/software.html).
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 2012
- Accession Number
- ADA567506
Entities
People
- Tryphon T. Georgiou
Organizations
- University of Minnesota