An Adaptive ARMA Spectral Estimator. Part 2.
Abstract
This paper presents a recursive algorithmic implementation of the prewindowed high performance method of ARMA spectral modeling as described in Part 1. This algorithm provides updates of the ARMA models optimal autoregressive coefficients (in actuality prediction errors) as each new data point becomes available. The algorithm is computationally fast in the sense that it requires O(p) multiplications and additions for each update. It is shown that this fast recursive algorithm may be implemented using a lattice filter arrangement, and it therefore exhibits several of the 'nice' properties associated with lattice type algorithms such as numerical robustness and good convergence properties. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1981
- Accession Number
- ADA101870
Entities
People
- James A. Cadzow
- Randolph L. Moses
Organizations
- Virginia Tech