Estimation of a Filtered Marked Poisson Process from Noisy Observations with Applications to Signal Processing
Abstract
The problem of estimating the arrival times, amplitudes, and phases of an unknown number of signals in noise is treated. The signals are assumed to have a common, known waveform. A Bayesian model using a Poisson prior for the arrival times is specified. and a real time algorithm for computing the posterior mode is developed. Alternatively, the procedure may be looked upon as a penalized likelihood estimator with a penalty team which is a generalized form of Akaike's Information Critierion. Simulation results are presented which show that this approach can improve over classical, linear methods.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1983
- Accession Number
- ADA127692
Entities
People
- Dennis D. Cox
- John E. Ehrenberg
Organizations
- University of Wisconsin–Madison