Viterbi Tracking of Randomly Phase-Modulated Data (and Related Topics).
Abstract
In this final report we formulate the problems of phase and frequency estimation as fixed interval smoothing problems. Forward dynamic programming algorithms are derived to find maximum posterior sequence estimates that pass likely sequences through the data. The algorithms extend the threshold usually associated with phase lock loops. The algorithms are applied to simultaneous phase estimation and data decoding in phase jitter channels. The resulting error probabilities are the lowest currently achievable. The results indicate that there are many problems in the domain of filtering and signal processing that can be profitably reformulated as sequence estimation problems. Our interest in likelihood leads us from frequency tracking to exact likelihood for autoregressive moving average (ARMA) data. Fast Kalman filtering algorithms are derived for constructing exact likelihood in multivariable ARMA models. Several interesting connections are established between Wold, Kolmogorov, Wiener, and Kalman representations of stationary time series. Ideas are proposed for associating spectra with linear transformations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 10, 1982
- Accession Number
- ADA118613
Entities
People
- Louis L. Louis L. Scharf
Organizations
- Colorado State University