Efficient Multichannel Autoregressive Modeling in Time and Frequency Domain.
Abstract
The single channel autoregressive lattice has been successfully applied to problems including speech analysis and recognition, spectral analysis and noise cancelling. More recently the two channel autoregressive (AR) lattice has been exploited for autoregressive moving average (ARMA) analysis of systems for modeling and identification. This dissertation considers the multichannel AR lattice when applied to ARMA systems analysis. Constraints on lattice parameters, based on the input output relations of the system under test, are developed. The lattice is redefined in terms of the frequency domain representation of the input data. This proves to be useful because it allows the input to be normalized so that the lattice yields a consistant set of parameters independent of the test source characteristics. Lastly the lattice is redefined in terms of correlations of the input signals. This results in a computationally and storage efficient lattice algorithm. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1982
- Accession Number
- ADA115752
Entities
People
- David J. Klich
Organizations
- Naval Postgraduate School