Graduate Handbook - 9. Faculty Research Projects

This work involves development of algorithms for automatic detection and identification of measurement error and network topology errors. Algorithms suitable for very large networks (several thousand nodes) are being investigated.

This work is involved with developing methods for efficiently solving the optimal powerflow problem for very large networks. The optimal powerflow problem involved determining the best dispatch of generation and voltage profile that minimizes some cost function related to the economy and/or security of operation of the network. The application of recently developed optimization methods to this problem is being studied.

Electric Power Network State Estimation with Partial Information

The current formulation of the electric power network problems assumes that measurement data are available through the network. This is not the case in practice since only limited data are available outside a control center's jurisdiction. A new formulation of the problem that incorporates the use of partial information, such as operating limits and generation dispatch, is needed.

Model based vision, automatic target recognition, industrial vision applications, and image fusion are subjects of concentrated work in the Convergent Technologies Center(CTC). New methods of Model decomposition and non-linear estimation have demonstrated promising performance in the general recognition problem.

Systems of multivariate polynomial equations arise in many optimization problems related to image fusion, target recognition, network theory and multivariate regression. Recent work has resulted in a new lower bound for the optimum value of the quadratic assignment problem (QAP), a combinatoric optimization problem that arises in many applications such as the allocation of processes in distributed computer system.

Design and implementation of computerized instructional modules employing color, animation, interaction, richly graphic interfaces and computation to assist students with the understanding of electrical engineering concepts and methods.

The design of a variety of parallel computer systems is being investigated. The Research is building on prior work into the design of data flow computers. These types of systems use the presence of data to trigger computations rather than the traditional control flow (the program counter) approach of the von Neumann type architecture.

This Research is concerned with the design of computer systems for real-time applications. The type of architectures currently being investigated are message-based loosely coupled parallel processor systems designed for hard deadline requirements. Related to this development are the implementation of novel scheduling algorithms and load balancing and process migration schemes.

The use of hardware description languages and computer aided engineering tools for the rapid prototyping and development of digital and computer systems is being investigated.

Capabilities to implement Analog integrated circuits need to be developed. Such structures are operational amplifiers, Analog-digital converters, switched capacitor filters are well understood. Special features of such devices are of interest in many areas of information and communication theory.

In VLSI design many software tools, such as layout editors, simulation and data conversion programs are needed. Existing programs developed at universities need to be improved. There is a need to develop more reliable logic simulators which is compatible with such design tools as MAGIC.

Current and voltage harmonics are most detrimental to the performance of power systems, causing overvoltages due to resonance, reduced life span of equipment due to additional losses, electromagnetic interference, error of energy meters, and instability of control systems. The following topics deal with the reduction of the total harmonic distortion in power networks: (1) Active Power Line Conditioning and Active Filters are used to generate "useful" harmonics that compensate the undesirable ones. (2) Use of Active Line Conditioner for Geomagnetic Induced Current Mitigation. This is a novel class of Circuits meant to create dc ampere turns that cancel the GIC. (3) Development of Unity PF Cycloconverters: Cycloconverters are notorious for generation of undesirable subharmonics and interharmonics.

The quantitative evaluation of the light flicker caused by Power Quality disturbance is a timely problem. The relation between the flicker caused by different types of light sources and the time variations and the distortion of voltage is the Main thrust of this Research.

Fifty-one 2kW adjustable nonlinear loads will be connected in a 15kV, 10MVA class feeder. This actual feeder will be used as an experimental laboratory to study filtering methods, electromagnetic compatibility, strong currents, statistical correlation and better simulation methods.

In order to assure the reliability of a numerical simulation, Research is currently conducted on automatic three-dimensional mesh generation for finite element/difference

Methods, the consistency and stability analyses of numerical methods, a posteriori error estimates for the numerical solution, and three-and/or four-dimensional )including time) visualization of electromagnetic fields.

As semiconductors and thus digital integrated circuits (IC) become faster, the inter-connections within the chips play an increasingly important part in circuit design. Research effort in this area con be divided into three subjects:

1. Semiconductor processing and device Modelings;
2. Coupling between active and passive components; and
3. Coupling between photonic and electronic components.

Since the parallel architecture presents a different method for performing calculations, it has become necessary to explore different classes of Partial Differential Equation (PDE) methods and assess their performance on parallel computers.

Fred J. Looft

The purpose of these studies is to develop an understanding of the stimulus encoder characteristics of tactile mechanoreceptors. Both linear and nonlinear systems analysis techniques are employed. During an experiment, electromechanical stimuli are applied to the skin overlying an isolated afferent. The action potentials resulting from the stimulation process are recorded for post-examination processing. Subsequently, the data is analyzed and used 2) to develop numerical Models for tactile receptors and b) to provide a basis for understanding how external events are reconstructed at the cognitive level.

This project targets the development of special purpose coils for organ specific image methodologies. Based on 3-D numerical analysis techniques new coil concepts are explored in an effort to enhance image resolution while reducing the coil size.

An inspection strategy is being developed to investigate the generation, propagation, and scattering of Ultrasound in a fiber reinforced composite undergoing curing. The Model predictions will be employed to optimize process control parameters such as temperature and pressure of the mOlding press.

Elastic anisotropy in single crystals has a profound effect upon the propagation of quantized elastic waves. Striking differences in the intensity of these waves can be predicted and experimentally observed along different chrystallographic axes. Energy flow can be strongly focused giving rise to enhanced heat conduction. This phonon-focusing effect is a direct result of the fact that the energy flow is, in general, not parallel to the propagation vector in elastically anisotropic crystals. All processes and interactions which involve phonons are thus affected by this phonon-focusing property. Recent Research in collaboration with A.G. Every, at the University of Witwatersrand in Johannesburg, has shown that piezoelectric stiffening of the elastic constants can profoundly change the nature of the group velocity surfaces in strongly piezoelectric crystals. In strongly elastically anisotropic crystals certain 3-wave interactions which conserve momentum and energy exhibit a reversal of the energy flow vector. Current Research involves the effects of phonon focusing on the theory of thermal conduction in single crystals at cryogenic temperatures.

The goal of this work is to guide IC design by connecting system-level performance to fundamental limits imposed by circuit-level considerations such as thermal and shot noise. Making this connection becomes more important as IC process improvements allow designers to push performance limits in speed, power and integration. Previous work has investigated sources of noise in an oscillator used as the time reference in a 155MHz data communication system. Future work will continue in the areas of Analog and mixed Analog/digital integrated circuits and systems.

Research in this area applies Analog techniques to improve performance in high speed, wide dynamic range electronic imaging systems. Applications include machine vision and adaptive optics.

This work is concerned with developing techniques for measuring and Modeling second order error sources in mixed signal circuits, for example, code-dependent noise in Analog-to-digital converters.

The FAA eventually plans to allow civil aircraft to use the Global Positioning System for all phases flight including precision approach and landing. One problem standing in the way is the fact that errors in the navigation information sent from the GPS satellites could result in a user calculating the incorrect estimate of their position. This Research focuses on developing, simulating, and evaluating algorithms that will detect these errors and notify users before air safety is compromised.

High-Performance Computer Systems

Research in high-performance computer systems is directed towards the development of a real-time, fault-tolerant, reconfigurable multiprocessing system.

Materials characterization using Ultrasound is explored using attenuation, which can be dependent on the wavelength, scatter size, and material constants. It has been demonstrated that power-laws are one appropriate expression for the scaling properties of attenuation and the grain-sized distribution in metals. Topics of interest are precision ultrasonic attenuation measurement, grain-size distribution determination using image processing techniques, and the theoretical development of the relationships between scatterer size and attenuation.

A new approach is taken to the study of ultrasonic scattering from rough surfaces using scaling properties. At large scales the surface often appears smooth, with a transition from the smooth properties to a rough topology as the surface is examined at a smaller scale. The surface roughness as a function of scale will be measured using the wavelength and angle dependence of a scattered ultrasonic wave. Numerical Modeling and theoretical development are being explored, using two-and three-dimensional surfaces.

Current image processing and robot vision projects include methods for determination of object orientation and identity from Images taken from any angle in three-dimensional space. Tensor-based and other non-iterative approaChes to the determination of geometric invariants are emphasized. Additional work is underway in determination of point correspondences, as well as in image synthesis.

In the power systems area, current work includes techniques of digital signal processing, spectral and statistical analysis, and real-time data acquisition and computation applied to measurement of power quality. In particular, the analysis, measurement, and characterization of current and voltage harmonics are studied together with the Modeling and prediction of system-wide effects on power quality.

Prof. Orr is active in the development of new undergraduate and graduate curricula in Electrical and Computer Engineering. Examples include a joint ECE-CS-MS level program in computer and communications networks, and first-year introductory Courses required in electrical engineering.

As Cryptography and data security gain importance, Researchers are looking for new mathematical functions which have "one-way" characteristic. That is, the function itself is easy to evaluate but the computation of its inverse is virtually impossible. Over the last few years, functions over elliptic curves have proved to be well suited and are applied in commercial systems. This project deals with the realization of a Cryptosystem utilizing hyperelliptic curves, which are functions which have recently become available from mathematical Research. The goal is the VLSI implementation of an entire Cryptographic product such as a smart card.

Many channel coding and data security schemes require arithmetic with numbers from a special type of algebraic structure, from so-called finite fields. This project's goal is the development of a processor with a very flexible and scalable arithmetic unit that allows fast arithmetic with those numbers. The new ALU to be developed would incorporate various new architecture designs which have been proposed over the last few years. The processor would have natural applications within communication environments such as data networks.

The data security aspects in networks such as wireless of ATM networks, may require specific Cryptographic approaChes. This project investigates algorithms which seem suited for given specific applications such as coding for video-on-demand or cellular phones.

The performance of a WIN system depends on effective radio propagation which requires Modeling and computer simulation of the radio propagation in the design stage. Statistical and deterministic Models are under investigation for radio propagation.

Based on radio propagation Models, CAD facilities for performance evaluation of radio modems are developed. Performance of various modems employing spread spectrum technology, adaptive equalizer or multi-sectored antenna are under investigation. Various modules for professional commercial software are developed to be used for design and analysis of the radio modems.

Wireless access methods are studied to determine the optimum method for sharing a common band among voice and data terminals. The purpose is to maximize the usage for al allocated bandwidth and create a fair etiquette or protocol to balance the usage among the voice and data terminals. Different network architectures are simulated in professional modulator software for the installation, design and analysis of the networks.

This Research involves determining absolute plaque reflectivity and Modeling the effects of plaque geometry and surface characteristics on the received signal from array transducers, for different insonifying fields. This will provide improved information on plaque characteristics.

Extraction of quantitative data from Pulse-echo measurements from surfaces and layered media requires that the scattered field from the reflecting interface be well understood. Current work is aimed at obtaining results for surfaces with various types of roughness.

Innovative biosensors and advanced chemometric techniques are being developed for accurate spectrophotometric measurements. An important goal of this Research is noninvasive spectrophotometric technique for continuous blood glucose measurement for diabetic individuals.

Biosensors are being investigated that make use of highly specific mechanisms for the recognition of complex biochemical species. The long-term biosensor Research focus is in the development of receptor-based biosensors utilizing microelectronic technology, molecular biology and receptor biochemistry.

Evolving cellular and personal communication networks require power control at the physical layer for interference management and energy saving. Current Research is directed toward autonomous power management and multiple access protocols for energy savings with uncompromised quality of service.

Many challenging problems stand in the way of a deeper understanding of network stability and communication system dynamics. Investigations are underway into the pathwise and statistical behavior of parallel processing systems, non-work-conserving queues, and optimistically synchronized computers.

Game theory is concerned with mathematical rePresentations of conflict and cooperation. Current Research is focused on the identification of core solutions, bounds on individual cost/benefit within the core, and corresponding geometric interpretation for general, n-player, cooperative scenarios.

Fundamental and largely unsolved problems in machine vision include the determination of the identity, location and orientation of 3-D objects in a scene. This Research involves the development of Model-based object recognition and orientation algorithms that are computationally efficient and employ compact Model structures. It also involves developing Models by which general machine vision problems may be cast into classical communication and information theoretic paradigms.

Image registration and fusion problems range from low-level processing for rapid automated inspection of micro-electronic circuits to large-scale, systematic approaChes to real-time multisensor fusion. This Research involves the solution of specific problems in the areas of image analysis, registration and fusion using a variety of strategies centered around the goal of developing rapid, analytically based algorithms.