Random Matrix Theoretic Approaches to Sensor Fusion for Sensing and Surveillance in Highly Cluttered Environments
Abstract
Powerful algorithms that are able to detect, estimate, and classify increasingly weaker signals buried in noise are a critical technological component of several important US Air Force technologies such as SAR, MIMO radar and hyperspectral imaging among others. Advances in VLSI are making sensors cheaper and easier to deploy in increasing numbers. What limits our ability to detect and discriminate weaker statistical signatures from statistical clutter is not the sensor count but algorithm-independent statistical limits associated with the finite number of snapshots over which the effects of clutter can be averaged out. In the work supported by the Young Investigator Award, we have characterized the fundamental limits of statistical estimation and detection for a variety of problems of direct relevance to the US Air Force. These are organized into two thrusts: Thrust 1: Characterization of fundamental limits and improved algorithms for detection, estimation and classification of matrix-valued signals buried in noise with missing data and, Thrust 2: Characterization of fundamental limits of and improved algorithms for transmission of energy through highly scattering random media.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 24, 2015
- Accession Number
- ADA621974
Entities
People
- Rajesh Nadakuditi
Organizations
- University of Michigan