Compressive Signal Processing
Abstract
In this project, we explored the theory and applications of compressive sensing (CS), in which non-adaptive, linear projections are used to acquire an efficient representation of a compressible signal or image directly using just a few measurements. The signal is then reconstructed by solving a tractable inverse problem. CS offers a fresh approach to framing and solving a number of timely and challenging problems in signal processing that relate directly to operational goals of the Air Force. Our approached was directed along three research thrusts. We first addressed the information scalability of CS. We applied CS principles to classification problems, showed that the CS measurement rate could be decreased with addition signal models, and investigated one-bit compressive sensing. Second, we improved on previous work in distributed compressive sensing. We created a joint manifold signal model, used graphical models to derive performance bounds on multi-sensor settings, and applied distributed CS concepts to a bearing estimation problem. Finally, we created a CS based radar framework and applied it to both 1-D and 2-D problems. Considering RF hardware, we proposed a compressive wide-band signal receiver, and we also used CS concepts for background subtraction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 30, 2010
- Accession Number
- ADA530830
Entities
People
- Richard G. Baraniuk
Organizations
- Rice University