Research

	Optics Laboratory © Donn Young

Our research is actively conducted by a diverse team of faculty at all academic ranks, including several world-renowned scholars, along with Master's and PhD students and research associates.

In recent years we have expanded our research focus, growing our department in both size and expertise and strengthening our research and graduate programs, especially in the area of Computer Engineering. We have increased our research productivity while continuing to improve our research and educational infrastructure.

Our programs are aided by strong support from industry leaders, many of whom are members of our Industry Advisory Board or adjunct research faculty.

Ongoing Research

Littoral Batlespace Sensor Fusion and Integration Framework
(Navy through Information and Systems Technology Research Center (IST-RC), 2007-08)

The Net-Centric environment currently being developed by the DoD, known as the Global Information Grid (GIG), provides a common fabric for systems to communicate. The GIG will provide the foundation for building an open architecture for automating sensor fusion. We propose to develop a sensor fusion framework for integrating the various parts of the sensor fusion process. It is a system of systems approach, riding on the GIG.  Our approach will allow data fusion algorithms to be easily connected with available data sources on the GIG (such as DCGS-N).  It will also provide the required interfaces and capabilities to support sensor fusion in the Net-Centric METOC battlespace.

Performance Evaluation Modeling for Multi-Sensor ATR and Tracking System
(Air Force Research Lab, SBIR Phase I subcontract through Intelligent Automation Inc., 2007-08)

In this project, we address both empirical and theoretical performance modeling approaches. We study the existing performance metrics for evaluating joint ATR and tracking algorithms. In addition to the popularly used metrics such as confusion matrix for ATR and root mean square estimation error for tracking, we also consider a new information theoretic measure to evaluate the joint system performance. The information theoretic metric intends to balance the ATR and tracking error instead of emphasizing only on one end, e.g., feature aided tracking or kinematic enhanced ATR. Extensive experimental and simulation studies will be categorized using the following scenarios (a) ATR alone; (b) tracking alone; (c) joint ATR and tracking. For ATR system, we focus on how the image quality indicators affect the classification accuracy and whether feature selection can improve the performance when one has limited training data. For ATR aided tracking, we will study how the classification on target ID can improve the tracking accuracy for closely spaced targets. For joint ATR and tracking, we propose to study certain tradeoff in classification and tracking accuracy using the information theoretic measure. The results also provide design guideline for sensor management where we assume the SAR/GMTI sensor can switch between sensing non-kinematic attribute and kinematic information of moving targets. Finally, we provide a few tasks on building reliable performance prediction models for multisensor system where track and ID fusion are needed. We highlight the major challenges in performance prediction for joint ATR and tracking system to persistently track evasive targets. We also propose to employ a game theoretic framework to schedule and assign sensors for sensing modes in realistic tracking environment.

Integrated Fusion and Routing System for UAV ISR Platforms
(STTR Phase I Air Force subcontract through Charles River Analytics, Inc., 2007-08)

As the quantity of available ISR sensor platforms increases, efficient cooperative routing of ISR assets for successful data collection becomes more difficult to coordinate. Additionally, the volume of information collected precludes comprehensive and timely human analysis. Automated routing algorithms and multi-source data fusion technologies are necessary to achieve decision superiority by exploiting the full capabilities of ISR assets. To explore possibilities for exploitation of knowledge gained by sensor fusion to the benefit of cooperative ISR platform routing algorithms, we propose to design and prototype an Integrated Fusion and Routing System (IFRS) for UAV ISR platforms. The proposed IFRS effort will examine how multi-source data fusion metrics can be used within a coevolutionary algorithm to dynamically improve cooperative ISR platform routing.

CSK Modulation for Underwater Communications,
(funded by SPAWAR through the Information and Systems Technology Research Center (IST-RC), 2007-08)

Information transmission underwater, without cables, will continue to be used as long as there are underwater vehicles or divers. Currently, the same techniques and digital modulation schemes developed for radio communications are being applied to the underwater acoustical channel.  Investigation of modulation schemes better suited to the unforgiving (noisy, changing, fading, multipath) underwater channel must be performed in order to increase the achievable range (separation between transmitter and receiver), quality (reduce the probability of error for a given signal power level), and speed (rate at which information may be transmitted) of these communications systems. This project concentrates in transmitting information contained in the rate of a chirp signal (carrier). In the simplest case, the slope of the linear frequency-vs-time chirp signal (up chirp or down chirp) contains the single information bit. The objective of this project is to investigate the performance of chirp-slope-keying for digital underwater acoustical communications. A small scale prototype of a CSK modulator/demodulator system will be designed and tested through computer simulations. Standard correlation and joint time-frequency transform methods will be used for demodulation of the chirp modulated signal at the receiver side. This spread spectrum system is expected to be more impervious to multipath, frequency fading, and other deleterious effects such as time delays and phase changes. Standard matched filter as well as joint time-frequency receivers (for example using the Wigner distribution and the Hough transform to detect the slope of the chirp) will be investigated to determine the performance improvement and computational burden of various receivers.

Cluster Expansion Project

The new 64 bit Linux cluster will have over 100 64bit processors, multi-terabyte of storage. The cluster is hosted in the College of Engineering and available to faculty for projects that require intensive computations. It was most recently used by Dr. Martin Guillot (Dept. of Mechanical Engineering) to run simulations for the Corps of Engineers to predict storm surge. The EE Department hopes to use the high-performance computing of its new cluster to improve the region's research capabilities in the wake of Katrina.

Non-invasive Real-time fatigue Monitoring System

Design of a non-intrusive real-time human fatigue detection and monitoring system for personnel working behind workstations and consoles. In many operational environments, any departure from alert human performance canhave extremely serious consequences. The system employs multi-modal vision systems for real-time online monitoring of physiological measures. The system is designed to allow for better risk management and personnel scheduling by command and control centers.

LEARN MORE

Read our flyer for Grad Students.