Bayesian sparse representation in colored noise: prewhitening vs joint estimation

In this paper, we consider the problem of designing sparse signal representation (SSR) amid colored noise. Two processing architectures are examined under a Bayesian framework: i) a two-stage processing with a prewhitening operation followed by SSR assuming a perfect white noise ii) a joint approach...

Full description

Saved in:
Bibliographic Details
Main Author: Bidon, Stephanie
Format: Conference Proceeding
Language:English
Subjects:
Bayes methods
Colored noise
Covariance matrices
Estimation
Numerical models
Radar
White noise
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we consider the problem of designing sparse signal representation (SSR) amid colored noise. Two processing architectures are examined under a Bayesian framework: i) a two-stage processing with a prewhitening operation followed by SSR assuming a perfect white noise ii) a joint approach estimating at the same time the sparse signal and the colored noise. Both approaches are compared; performance is numerically studied in case of conventional radar scenarios. Results show that the joint algorithm outperforms to some extent the prewhitened approach but at the expense of a higher complexity.
ISSN:2375-5318
DOI:10.1109/RADAR.2019.8835657