Current

Fall 2008

Archive

Spring 2008

Fall 2007

Spring 2007

Fall 2006

Spring 2006

Fall 2005

Spring 2005

Fall 2004

Spring 2004

Fall 2003

Spring 2003

2002 and Older

RFID: The Cow's Perspective

12:30 p.m. Tuesday, Jan. 23, 2007

in EE 125

by Dustin Vaselaar

Abstract

An overview of Passive UHF RFID will be discussed with respect to its design and application in cattle industry. Some of the basics behind RFID will be reviewed in order to serve as launching pad for examining some of the technical details of RFID hardware, communication channel modeling, and tag performance measurement with an emphasis on livestock applications. A brief status report will be given on the USDA sponsored "Cattle Tag" project which has funded this work.

Top

Dynamic Modeling of Tunable Gain Switched Dye Lasers

12:30 p.m. Tuesday, March 6, 2007

in EE 125

by Dr. Ivan Lima

NDSU
Fargo, ND

Abstract

Tunable dye lasers continue to be the choice for many applications including cancer diagnostics and other medical treatment due to their unique characteristics, including a wide tunable range in the visible and in the infrared. The dynamics of sub-nanosecond gain switched dye lasers can be accurately modeled using the rate equations. The rate equations describe the pump intensity, the laser intensity, and the density of molecules in the different allowed energy levels in the lasing medium. We show how to implement a new numerical method that decreases the computational time to model dye lasers by as much as two orders of magnitude when compared with standard finite differences method. We validate the method that we propose by comparison with experimental results obtained from a sub-nanosecond gain switched tunable dye laser.

Top

Microwave Amplifier Design

7:30 a.m. Friday, March 9, 2007

in EE 217

by Zakir Hussain

NDSU
Fargo, ND

Abstract

This work presents electronics amplifier design procedures for the high-frequency (HF) range up to frequencies in the microwave range. Electrical engineering textbooks typically focus on amplifier design and analysis for a particular range of frequencies. For example, introductory electronics textbooks only present amplifier design techniques that are suitable for frequencies up to a few megahertz. On the other hand, microwave engineering textbooks provide amplifier design techniques that are suitable only for microwave frequencies. The objective of this paper is to present amplifier design methods that are useful for the full range of frequencies commonly encountered by the radio-frequency (RF) or microwave design engineer.

This presentation focuses on the single stage BJT amplifier design starting with basic transistor amplifier theory. For high-frequency amplifiers, issues such as stability, impedance matching, and the effect of parasitics, are presented in terms of their circuit elements and also from the perspective of two-port parameters. Examples of amplifier designs for the HF band and for microwave frequencies are provided. These designs were studied using a variety of software programs and then constructed and tested. Their measured characteristics are compared to the analytical and simulated results.

Top

Investigation of the Central Tendency Measure as a Parameter for Brain-Computer Interface Thought Discrimination

12:30 p.m. Tuesday, March 20, 2007

in EE 125

by Adam Black

NDSU
Fargo, ND

Abstract

In the past few decades, chaos theory has become a growing discipline in nonlinear dynamics. More recently, it has been shown to have useful applications in biosignal analysis. A particular nonlinear parameter, the central tendency measure (CTM), can help quantify the degree of variability in brainwave data. This presentation will introduce the CTM and describe its effectiveness when used as a feature to discriminate between various mental thoughts in a brain-computer interface application.

Top

Impedance-based Structural Health Monitoring Circuit Design for Piezoelectric Structures

3:00 p.m. Tuesday, March 20, 2007

in Sudro 21

by Shirui Wang

NDSU
Fargo, ND

Abstract

Systems and structures monitoring their own structural integrity have become an active field. Traditional methods use Non-Destructive Evaluation (NDE) and Non-Destructive Testing (NDT). However, in order to lower the inspection costs, the research on intelligent material systems is becoming more active. This technology has practical applications in many areas such as aerospace systems, machine parts, and civil buildings.

One method uses the piezoelectric impedance-based structural health monitoring technique, which utilizes a piezoelectric patch attached to a structure and measures its impedance within a certain frequency range. The frequency is maintained in the kHz range for optimum sensitivity and coverage in damage detection. Piezoelectric materials are used as both actuators and sensors. Due to coupling between electrical impedance and mechanical impedance, analyzing electrical impedance variations can indicate physical changes in the host structure. Most methods use expensive impedance analyzers.

Based on the idea of a bridge circuit, a new electronic circuit to realize electrical impedance monitoring is presented. The proposed circuit generates a frequency sweep from 53 kHz to 164 kHz and measures the electrical impedance modulus relative value of a piezoelectric patch. Costs of this method are much lower than traditional methods. An integrated circuit of self power circuit, impedance measuring circuit and wireless communication circuit may be realized in the future.

Top

Localization-Based Handoff for 802.11 WLAN

2:30 p.m. Thursday, March 22, 2007

in FLC 415A

by Hareesh Khattri

NDSU
Fargo, ND

Abstract

Reducing the handoff delay experienced by a WLAN mobile station roaming between access points is important to improve the performance of WLAN for real time applications such as video streaming, which require less than 50ms handoff delays. The current standard for WLAN roaming is IEEE 802.11f.

A new predictive model is used for localization of mobile clients in a wireless LAN. One application of this method is to reduce the handoff delay in 802.11 WLAN. Localization of WLAN devices is done using a pre-stored signal strength distribution map of the coverage area and instantaneous signal strength measurements. A new probabilistic model utilizing Bayesian inference and Expectation Maximization techniques is used to improve location tracking. Syncscan method is used to continuously monitor signal strength values.

Implementation and performance testing of the new 802.11i security standard was also done. This study shows that the new security standard increases handoff re-association and authentication timings significantly.

The proposed localization method was implemented and tested on a real WLAN environment. The average localization accuracy is shown to be 88.37% and shows improvement over existing localization methods. This method can accurately track the movement of a WLAN user moving with velocity less than 12 ft/sec.

Top

The Divergence of Software Programming Models from Hardware Semantics

12:30 p.m. Tuesday, April 3, 2007

in EE 125

by Tim Brookins

Microsoft
Fargo, ND

Abstract

The abstraction level of mainstream programming languages has closely mimicked the semantic level of the CPU for most of computer science history. CPU's had (conditional) branch instructions and programming languages had "if" statements. CPUs had jump instructions and programming languages had "for" loops. CPU's had single cores and mainstream programming languages had a sequential control flow model.

Now we enter an age where clock frequency constraints have altered the semantics of the modern CPU. Multi-core is the future of hardware and programming languages need to react. To follow the historical model, programming languages would simply change to reflect this new hardware model, shifting to a parallel programming paradigm. But parallel programming can be extremely difficult for mainstream programmers to master and there is a huge body of existing software which cannot be easily adapted to this new paradigm. This seminar will explore these challenges.

Top

Solar Power and Maximum Power-Point Tracking

12:30 p.m. Tuesday, April 17, 2007

in EE 125

by Juline Shoeb

NDSU
Fargo, ND

Abstract

Over the last few decades, a great deal of research has been conducted in the fields of renewable energy sources. Among the renewable energy sources, solar energy is considered to be one of the most useful energy sources due to its clean nature and possibility of quiet harvesting. The main drawbacks of a solar cell are high cost, and nonlinear voltage-current characteristic which complicates the harvesting process.

The most practical and economic way of operating a given Photovoltaic (PV) system is to develop an effective algorithm that can be implemented with the help of simple electrical circuits and controllers to track the maximum power point of the system so as to extract as much electrical power as possible. Most of the existing maximum powerpoint tracking (MPPT) algorithms are very successful in tracking the maximum power points of a given PV system. However, the implementation of these algorithms results in cost-inefficient and very complicated PV systems. The seminar will investigate all of the existing MPPT methods to determine the reasons for low cost-efficiency and the technical complications associated with the implementation of these algorithms. The scope of this seminar also includes the development of a new cost-efficient MPPT algorithm for solar panels along with the application of PV systems for lighting with power light emitting diodes (LEDs).

Top

Bit-Level Pipelined Machines Using Refinement

3:00 p.m. Friday, April 13, 2007

in Sudro 22

by Sudarshan Srinivasan

Georgia Institute of Technology
Atlanta, Georgia

Abstract

Functional verification of microprocessors is a critical problem as processor designs are extremely complex, deeply pipelined, and highly optimized. Existing approaches for verifying bit-level pipelined machines--machine models that describe the pipelined behavior of microprocessors--are based on theorem proving and require extraordinary expert user effort.

We present a novel, efficient, and scalable approach for the verification of bit-level pipelined machines using refinement. Proving refinement guarantees that the pipelined machine behaves like its instruction set architecture. The verification approach uses a theorem proving system such as ACL2 to reduce the bit-level verification problem to a term-level problem, an abstraction of the bit-level problem. The term-level problem can then be solved in an automatic and efficient manner using several refinement-based methods that we have developed that use decision procedures. We demonstrate the efficiency of our approach by applying it to verify a complex Intel XScale inspired processor model that implements 593 instructions.

Biography

Sudarshan Srinivasan is currently pursuing his Ph.D. in Electrical and Computer Engineering at the Georgia Institute of Technology. He received an M.S. in Electrical and Computer Engineering from the Georgia Institute of Technology in 2003 and a B.E. in Electrical and Electronics Engineering from the University of Madras in 2001. His research interests are in Formal Verification, Hardware Validation, Computer Architecture, and Computer-Aided Design of Digital Systems.

His current research focus is in the development and application of Formal Verification methods to hardware systems

Top

Robust Nanoscale Circuit Design Considering Manufacturability

3:00 p.m. Monday, April 16, 2007

in Sudro 22

by Shiyan Hu

Texas A&M University
College Station, Texas

Abstract

As VLSI technology enters the nanoscale regime, demands for minimum feature sizes have outpaced the advances in lithography hardware solutions. This imposes a great challenge on manufacturing reliability. In current lithography technology, 193nm wavelength is used to print 45nm features. As a result, images on wafer have significant mismatches from mask layouts, and timing and power of circuits become increasingly unpredictable. It is imperative to consider the manufacturability issue during layout design (i.e., design for manufacturability) such that the chip design can be reliably fabricated.

My work in this emerging research area develops new techniques that address the challenge of achieving high manufacturing yield with low design cost. In the talk, I will describe the current design and manufacturing flows, the impact of layouts on printability, and how to improve the robustness of design by the design for manufacturability techniques.

Biography

Shiyan Hu is a Ph.D. candidate in the Department of Electrical and Computer Engineering at Texas A&M University. His research interests are in the area of design automation for robust and low-power nanoscale circuits.

Top

Robustness Analysis of Inter-area Oscillations Control in Power Systems

3:00 p.m. Friday, April 20, 2007

in Sudro 22

by Dr. Lingling Fan

Midwest Independent Transmission System Operator
St. Paul, MN

Abstract

Inter-area oscillations are common problem in most interconnected power systems due to long distance transmission. Damping control through Flexible AC Transmission System (FACTs) is a practice adopted. The best control signals for inter-area oscillations are usually "global signals". Through wide area measurement devices, these control signals are available. However significant time delay becomes a concern for controllers. Time delay weakens the system by inducing oscillations.

Quantifying the impact of time delay can be done by uncertainty modeling and small gain theorem. Modeling time delay as normalized coprime factor uncertainty, finding the minimum norm of the uncertainty, and further comparing with robust stability margin, we can tell if the controller can tolerate the time delay. It is found that a properly designed FACTS damping controller can achieve damping performance very well at various power transfer levels. In terms of robust stability, the controller can tolerate a wider range of time delay at lower power transfer level.

Biography

Lingling Fan received her bachelor's master's degrees in Electrical Engineering from Southeast University (Nanjing, China) in 1994 and 1997. She received a Ph.D. in Electrical Engineering from West Virginia University in 2001. She has worked for Midwest Independent Transmission System Operator since 2001. Her research interests include power system dynamics and control, power system reliability and economics.

Top

Intelligent Hierarchical Control of a Multimachine Power System

3:00 p.m. Monday, April 23, 2007

in Sudro 22

by Dr. Salman Mohagheghi

Georgia Institute of Technology
Atlanta, Georgia

Abstract

The traditional approach of controlling a power system using local control agents has several drawbacks, such as sub-optimality of the local controllers over the wide range of system's operating conditions, as well as unwanted interactions between the controllers. These problems can be solved by using a hierarchical control scheme that has information on the overall performance of the power system and its local controllers. The hierarchical controller can provide auxiliary control signals to the controllable devices such as generators and/or FACTS devices. However, the classical control methods often fail to provide a very effective solution due to the complexities and the nonlinearities of the multi-input multi-output non-stationary power system. The aim of this presentation is to introduce the concept of intelligent hierarchical control for a multimachine power system as an alternative for providing multi-level control in the power grid.

Biography

Salman Mohagheghi received the B.Eng. from University of Tehran, Iran in 1999 and M.Sc. from Sharif University of Technology, Tehran, Iran in 2001, both in Power Electrical Engineering. In 2006 he graduated with PhD in Electrical Engineering from Georgia Institute of Technology, where he is currently employed as a postdoctoral fellow. His research focuses on wide area control in power systems, protective relaying and distributed state estimation.

Top

Model Based Analysis and Design of Distributed Real-time Embedded Systems

3:00 p.m. Friday, April 27, 2007

in Sudro 22

by Dr. Sherif Abdelwahed

Vanderbilt University
Nashville, Tennessee

Abstract

Distributed real-time embedded (DRE) systems, including intelligent transportation, automated inventory management, command and control systems, and avionics mission computing, increasingly run in open environments, such as network-centric systems of systems. This emerging operation setting introduces new challenges for DRE system developers, such as managing the system performance under uncertain operating conditions, validating key-characteristic of the system behavior, and choosing the right design alternatives before committing to a specific platform or platform configuration. Model-based technologies help address these issues by enabling design-time analysis and providing means to automate the development, deployment, run-time adaptation, and integration of DRE systems.

This presentation introduces two promising research directions focusing on model-based design technologies, namely, the development of self-managing DRE systems using automatic control and system-theoretic concepts, and the application of model-checking techniques to verify safety properties of DRE systems. Results of this work show that model-based techniques can be effectively applied to manage, predict, and verify the complex event-driven behavior of DRE systems. The presentation introduces several practical implementation of this model-based technology and discusses related future research directions.

Biography

Dr. Abdelwahed received his Ph.D. degree in Electrical and Computer Engineering from the University of Toronto, Canada, in 2002. During 2000-2001, he was a research scientist with the system diagnosis group at the Rockwell Scientific Company. Since 2001 he has been with the Department of Electrical Engineering and Computer Science at Vanderbilt University as a Research Assistant Professor. He conducts research on model-based design and analysis of self-managing computation systems. His research interests also include modeling and analysis of distributed real-time systems, automated verification, fault diagnosis techniques, and model-integrated computing. He is currently working on several DARPA, NASA, and NSF funded projects developing model-based techniques for fault-adaptive performance management of distributed real-time embedded systems. Dr. Abdelwahed has more than 60 publications and is a senior member of the IEEE.

Top

Modified Hilbert-Huang Transform for Power System Applications

3:00 p.m. Monday, April 30, 2007

in Sudro 22

by Dr. Nilanjan Senroy

Florida State University
Tallahassee, FL

Abstract

The Hilbert-Huang transform has recently been proposed as an adaptive signal processing technique to study non-stationary/non-linear signals encountered in diverse fields ranging from electrical engineering, geophysics, climatology and oceanology, and biomedical applications. In power systems, it has been applied to examine inter-area oscillatory modes, torsional shaft signal attributes in sub-synchronous resonance studies and to recognize transient power quality events like voltage dips. The first part of this technique involves empirically decomposing a complicated signal into its constituent modes of oscillation. Hilbert transform on these modes yields their analytic forms, from which their instantaneous amplitudes and frequencies are computed.

In this presentation, the fundamental assumptions of the original Hilbert-Huang method are revisited, and some modifications are proposed to enhance its application as a 2nd order filter in power system studies. Two applications of the modified Hilbert-Huang transform are presented to demonstrate its accuracy and adaptability.

Biography

Dr. Nilanjan Senroy is a post-doctoral associate at the Center for Advanced Power Systems, Florida State University, where he is working on power quality issues onboard a notional all-electric ship. He holds a Ph.D. from Arizona State University. His research interests include wide-area stability of power systems, modeling and simulation, distributed generation and signal processing applications in power systems.

Top

The Effect of Heart Rate, Preload, and Afterload on the Viscoelastic Properties of the Bovine Myocardium

by Bruce Wheeler

NDSU
Fargo, ND

Abstract

Experiments were performed to test the hypothesis that visco-elastic properties of the bovine myocardium are independent of heart rate (HR), preload (PL) and afterload (AL). Left ventricular pressure and aortic flow (AoF) waveforms were recorded in ten bovines. At different paced heart rates, an inferior vena caval occlusion (IVC) was used to reduce PL, then the IVC was released and simultaneously the aorta was clamped to increase AL. Equivalent left ventricular pressure waveform pairs consisting of an ejecting waveform (denoted as LVP) and isovolumic waveform (denoted as hydromotive pressure, HMP) were selected according to specified criteria resulting in 448 equivalent waveform pairs from seven bovine. From the selected waveform pairs and corresponding aortic flow waveforms, the visco-elastic properties (k and ε₁) were estimated by: HMP = LVP + εV_EJ + k*LVP*AoF. Here ε₁ is the parallel elastance, k is the myocardial friction, and V_EJ is the integral of AoF over ejection. The parameters k and ε₁ were calculated for 448 matched waveform pairs (average k = 0.0006 ±0.000016 s*mL^-1, and average ε₁ = 0.6406 ±0.0306 mmHg*mL^-1 (α = 0.05). The visco-elastic parameters (k and ε₁) did not exhibit any clear or predictable dependence on HR, PL, and Al.

Top

Classification of EEG Using Support Vector Machines

by Monobina Bhowmick

NDSU
Fargo, ND

Abstract

A key aspect of brain computer interface (BCI) systems is the ability to classify various thoughts or intentions. Support Vector Machines (SVMs) are a novel classification technique based on statistical learning theory. A Matlab-based toolbox was used to evaluate various aspects of SVMs when applied to electroencephalographic (EEG) data during actual and imagined hand movement. Four types of SVM kernels were compared: linear, polynomial, spline and radial basis function. Additionally, parameters that affect the number of support vectors and various normalization methods were investigated. Overall, all kernels except the linear case achieved zero percent error during the training phase. The test phase results also showed that the linear kernel performed below that of the nonlinear kernels. For all kernels, the hand movement trials were classified better than the imagined hand movement trials. The normalization results varied but it is generally considered good practice to apply a normalization method.

Top