The MERIT Summer Research Program
  The A. James Clark School of Engineering
  University of Maryland Home

Rite '00 Project Descriptions

RITE Site Project Descriptions: Summer 2000

1. Advanced Components for Mobile and Optical Communication Systems
Prof. A. Iliadis

In this project Micro-Electro-Mechanical Systems (MEMS) technology is employed to miniaturize antennas for mobile communication platforms. Antennas will be modeled and their design optimized for sizes suitable for monolithic integration with personal communication cards and hand held units. Emphasis is placed on developing new geometries and using novel materials deposited by Pulsed Laser Deposition (PLD) such as oxides and metallic compounds, both for the efficient propagation and containment (shielding, isolation) of the signals. MEMS technology and PLD deposition will be applied for the processing and fabrication of these "chip" form antennas.

Current mobile communication hand held units use an analog front-end receiver/transmitter amplifier design that down-shifts (or up-shifts for emission) to an intermediate frequency compatible to the digital signal processor (DSP) unit that then processes the in-coming (or out-going) signals. In this project we study the conditions under which the DSP can be brought to the front end of the system, the specific requirements of the system in terms of shielding, power, and frequency response.

Students will have the opportunity to learn the intricate details of antenna design, use software to model and simulate wave propagation, develop processing and fabrication skills, and understand the properties of novel materials critical to the development of the next generation of microelectronics systems. Furthermore, students will examine and evaluate different circuit design approaches, introduce novel absorbing and shielding materials, and interact with industry.

2. Computational Models of the Human Auditory System
Prof. S. Shamma

Over the last few years, Prof. Shamma's group has conducted summer projects with several undergraduates dealing with how the auditory system processes sound, from the ear up to the brain. Projects suitable for RITE participants are those involving the use of an extensive library of auditory signal processing algorithms and graphics dedicated to speech and music analysis and editing. Developed in MATLAB at the University, this package includes help and graphics features, as well as sophisticated interfaces and interactive capabilities that make it particularly easy to use by undergraduates. In using this software, students focus initially on the basic description of the processing stages and their underlying concepts, treating the computational elements as black boxes. They learn first how to perform various useful operations on sound signals such as transformations between time and frequency domains, conditioning, morphing sentences into each other, and altering voices. Once they acquire familiarity and facility with the package, they move on to examine the underlying principles behind each command and operation. In the process, students learn how to use MATLAB commands, various linear system concepts and operations, filtering, and some nonlinear transformations. In many cases, particularly interested and bright students may go on to create new algorithms (albeit simple), to modify existing ones, or to invent new applications ranging from recognition of musical instruments to the synthesis of music and speech.

The projects described above are fully supported by extensive computational and laboratory facilities at the Neural Systems Laboratory.
Last summer, Professor Shamma supervised three RITE students who received the best project award during the MERIT fair.

3. Computer Recognition of Faces
Prof. Rama Chellappa

Computer recognition of faces has applications in access control, ATM machines and in human/computer interfaces. In the project, the student(s) will develop global principal component analysis/linear discriminant analysis or feature-based approaches for verifying the presence/absence of a human and then determine the human's gender, identity and emotions. Using images acquired by a camera mounted on a PC, the algorithms will be able to produce an electronic log of the human's arrival/departure record. Emphasis will be placed on real-time implementations of existing algorithms in Professor Chellappa's group.

4. Design of High-Power Vacuum Electronic Amplifiers for Communication Systems
Prof. Wes Lawson

The wireless communication industry is growing today at an enormous rate both in terms of the total number of users and in the number and types of services that are available. In order to meet these increasing needs, a number of institutions are looking at the prospects of high frequency communication systems. One limitation of high frequency communications is the lack of roadband high-power sources. Gyrotrons are devices that have demonstrated the ability to produce record-level powers at millimeter-wave frequencies for narrowband applications. In this project, you will learn about the requirements for communication sources and about the operation of microwave tubes. Then you will design gyrotron tubes and related system hardware that have broadband capabilities. Finally, you will evaluate the suitability of your microwave systems for satellite and other communication applications.

5. Domain-Specific Compilation Techniques for Multimedia Software Implementation
Prof. S. Bhattacharyya

Over the past several years, high-level, domain-specific design automation tools have become increasingly important for the development of multimedia applications. Such tools employ innovative programming models and compilation techniques that exploit the distinguishing computational characteristics of multimedia applications. The numerous commercial tools in this category include the Signal Processing Worksystem from Cadence, COSSAP from Synopsys, ADS from Hewlett-Packard, and DSP Canvas from Angeles Design Systems. Parallel processing has been one of the most active research topics during recent years, and offers great potential for multimedia applications. Among parallel architectures, those that employ different types of processing units (heterogeneous multiprocessors) have become especially promising in embedded multimedia systems. The exploitation of such architectures requires effective partitioning and scheduling strategies that account for component heterogeneity at all levels. In the context of multimedia design automation technology, this project will explore compilation techniques for systematically mapping applications onto heterogeneous multiprocessor architectures. The project will include extensive development and use of simulators at various levels of abstraction to expose insights into the underlying problems and to validate the results that are developed. In addition to providing valuable intuition about multiprocessor implementation trade-offs, the development and use of such simulation tools will introduce students to research issues in multimedia software implementation

6. GPS Based Location Determination
Prof. P. S. Krishnaprasad

The nucleus of a GPS (global positioning laboratory) has been set up within the Intelligent Servosystems Laboratory, a joint facility of the Electrical and Computer Engineering Department and the Institute for Systems Research. This is the outgrowth of projects and courses on GPS Theory and Technology over the period 1997-1999.

At the present time the lab includes a mobile remote controlled platform (actually a meter long radio-controlled car equipped with an internal combustion engine), with on-board Novatel GPS receiver, and a Novatel base station receiver, and telemetry channel enabling differential, carrier-sensed GPS positioning to an accuracy of 2 cm, based on commercial and hand-crafted estimation algorithms. The new algorithms are being developed by a Ph.D student and an undergraduate student is involved in ongoing experimental work.

GPS represents a remarkable integration of several major advances in the field of electrical engineering--from satellite communication, coding, estimation of signals in noise, to real-time signal processing via fast, low power circuitry on portable devices--providing us the capability to determine with increasing precision and accuracy, the location and speed of a GPS-equipped platform practically anywhere on earth. Emerging practical applications of this technology range from high accuracy surveying for geographical information systems (GIS), high-precision geodesy for earth science, to accurate navigation and flight control of aircraft, personal and commercial land vehicles, autonomous farm equipment, autonomous mobile robotics, to name a few. Bringing this technology to the attention of undergraduate students would be useful in strengthening the empirical base and technical preparedness of a typical student.

It is expected that initially, work in the proposed project would emphasize a view of the integration of subject matter inherent to GPS, demonstrating how systems concepts, such as filtering algorithms (e.g. Kalman filters and various nonlinear filters), mathematical descriptions of orbital mechanics of satellites, and differential or interferometric techniques, and phase tracking techniques contribute to the advances in precision and accuracy achievable via GPS. Electrophysical aspects with an impact on a variety of noise/error phenomena (ionospheric distortions, foliage-induced loss-of-lock and multipath) have to be dealt with properly in a successful GPS system. The project will also provide opportunities for studying applications of GPS to motion control, and in this context also investigate integration of a variety of other sensor technologies (example inertial and optical/acoustic proximity sensors) with GPS for this purpose.
It is expected that initially, work in the proposed project would emphasize a view of the integration of subject matter inherent to GPS, demonstrating how systems concepts, such as filtering algorithms (e.g. Kalman filters and various nonlinear filters), mathematical descriptions of orbital mechanics of satellites, and differential or interferometric techniques, and phase tracking techniques contribute to the advances in precision and accuracy achievable via GPS. Electrophysical aspects with an impact on a variety of noise/error phenomena (ionospheric distortions, foliage-induced loss-of-lock and multipath) have to be dealt with properly in a successful GPS system. The project will also provide opportunities for studying applications of GPS to motion control, and in this context also investigate integration of a variety of other sensor technologies (example inertial and optical/acoustic proximity sensors) with GPS for this purpose.
Students selected under the RITE site program will be able to pursue software development in this arena (with interesting challenges in integer optimization for phase ambiguity resolution, cycle slip detection and mode switching in satellite tracking). A primary goal of the project will be for students to learn the principles and carry out implementations of motion determination algorithms based on filtering and estimation theory and gain a better appreciation of issues in signal processing influenced by noise characteristics associated to the GPS system.

7. Installation/Configuration and Performance Analysis of an IPv6 Testbed
Prof. John Baras

The Internet Engineering Task Force (IETF) has produced specifications (RFC 1752, 1883, 1886, 1971, 1993, etc.) that define the next-generation IP protocol known as "IPng," or "IPv6." IPv6 is both a near-term and long-range concern for network owners and service providers. IPv6 products have already come to market; on the other hand, IPv6 development work will likely continue well into the next decade. Though it is based on much-needed enhancements to IPv4 standards, IPv6 should be viewed as a new protocol that will provide a firmer base for the continued growth of today's Internet work.

This project involves setting up and IPv6 test-bed in the Hybrid Networks Lab. (HNL). The test-bed will be linux based. This will involve investigating the emerging IPv6 standards and evaluating the various distributions available. The test-bed will involved at least 2 linux PC's. The HNL IPv6 test-bed must be integrated with the existing 6bone (set of known Ipv6 capable machines in the Internet). This project will benefit the HNL, as it will provide a running IPv6 test-bed to the lab for future use in various projects. This project will benefit the students as it will make them familiar with an emerging standard in the Internet community. It will also familiarize them with various issues related to IPv6, its deployment, and its integration with IPv4

Preferred skills:
--Familiarity with Linux, installation/administration/configuration
--Familiarity with networking concepts
-- Familiarity with TCP/IP

8. Line of Sight Optical Communications Links
Prof. C. C. Davis

With the crowding of the electromagnetic spectrum with RF and microwave emissions and the need for regulated spectrum allocation, there has been recent renewed interest in laser-based line-of-sight optical links for both military and industrial uses. These links offer high data rates (>1Gb/s) over modest ranges of a few kilometers, as well as intrinsic security. They operate over modest ranges in all but the most severe conditions of atmospheric obscuration. The project goal is to parameterize the performance of such links so that inexpensive, high performance systems can be deployed for industrial and military uses. As an adjunct, a real, multi-kilometer link testbed will be built for long term studies of atmospheric turbulence and its effects on line-of-sight links.

A number of ongoing experiments are ideal for undergraduate participation. The current test setup uses a water-filled, heated tube to mimic the effects of a kilometer-length atmospheric path that can generate turbulence conditions ranging from weak to strong. The type of experience that RITE participants would obtain in the laboratory is best illustrated by the work of Julie Freidlin, an former undergraduate member of the research team. Julie designed the mounts to hold the turbulence tube and supervised their manufacture. She also designed and built a variable gain optical receiver and is currently writing code to collect and analyze data.

During the first RITE Site program, Summer 1998 Professor Davis worked with two undergraduates (Ronald de Guzman, GWU, and David Wenzel, University of Virginia) on "Characterization of an Optical Communications Channel." David continued his work with Professor Davis during the 1998-99 academic year and is currently a graduate student in the Department of Electrical and Computer Engineering at the University of Maryland.

9. Multi-Camera Tracking of People
Prof. Rama Chellappa

In this project, the student will use images collected by cameras in the Keck laboratory to track a person or persons as they come in view of the cameras and produce continuous trajectories of persons moving around. The emphasis is on developing real-time algorithms using recursive filters such as the Kalman filter or other non-linear filters. The approach will be generalized to include person's identities as a component of the state vector that is propagated.

10. Network Security for Emerging Broadband Networks
Dr. Patrick Dowd

As the world becomes more connected, the need for advancement in network security has become crucial. Network security mechanisms need to be developed to keep pace with this very exciting environment. It is not just the advancement in terms of network speed - it is also the integration of voice, video and data on the same network. Voice over IP is extremely important and is expected to become a dominant vehicle for voice transport. This project will develop security methods to protect voice traffic in an IP-based (internet) like environment, and will also study the interface between the SS7 public switched telephone network (PSTN) to the IP network. The goal is to develop mechanisms for high speed fire walling and network intrusion detection in this context. This project will have both experimental and theoretical components. The experimental testbed will be ATDnet/MONET - a multi-gigabit WDM research network that is located in the Washington DC area. The research project will build from a current research project on high speed firewalling for broadband networks and will also include the participation of a major telecommunications company.

11. Real-Time Operating System Support for a Wireless Network of Sensors and Actuators
Prof. David Stewart

Students in the MERIT '99 program built the real-time operating system (RTOS) support for point-to-point wireless communication between sensors and actuators in a distributed control system. Device drivers and prototype hardware were built, to demonstrate both IR and RF transmissions, and the RTOS's ability to provide a transparent interface to each. As a next step for MERIT '00, we intend on using broadcast messages, so that there may be multiple senders and multiple receivers of the data. The goal of this project is to design the system software that enables application designers to incorporate remote wireless sensors and actuators in a transparent fashion, allowing them to design software as though all these sensors and actuators are local. The system software would take care of all of the real-time scheduling, synchronization, and error detection and handling issues to guarantee that all necessary communication occurs in a timely fashion.

12. Reliable Multicast Testbed: The Effects of Local Feedback on Multicast Data Distribution Protocols' Performance
Prof. John Baras

The Adaptive File Distribution Protocol (AFDP) addresses the need for efficient and reliable group communication. The protocol is built on top of UDP, and uses a rate-based flow control mechanism to provide efficient and reliable file distribution. AFDP uses the publishing metaphor: any machine receiving files is called a subscriber and any machine sending files is called a publisher. One special subscriber is designated as the secretary; this machine is responsible for managing the group membership and authorizing publishers. AFDP dynamically selects between 3 different transfer modes: unicast, multicast (on hosts that support it), and broadcast. This dynamic selection of transfer modes is very useful for tuning network and host loading, throughput, and for dealing with hosts and networks that don't support multicast. If the number of hosts in the group is very small, it is often best to unicast to each subscriber. As the group grows, AFDP may switch to either multicast (if all the hosts and networks support it) or broadcast (if all the hosts are on the same LAN). (Reference paper: Why Use a Fishing Line When You Have a Net? An Adaptive Multicast Data Distribution Protocol by J. Cooperstock and S.Kotsopoulos. In Proceedings of Usenix '96.)

This project will study the applicability of the AFDP protocol to a DBS environment and how local recovery can be used to enhance the operation of this protocol over a DBS system. The test-bed will be setup in the HNL using the satellite channel emulator to provide the DBS environment. The student will be required to understand the working of this protocol, and then configure/modify/change things as required to enhance its performance for the emulated DBS system in the HNL. This project will also study how the current protocol can be enhanced by providing a system for Local Recovery of lost data.

Preferred Skills:
--Ability to install/configure software
--Some programming ability C/C++
--Ability to setup and configure workstations for the testbed
--Understanding of basic multicasting principles

13. Studies of Vacuum Electronic Amplifiers for Chaotic Communications
Prof. T. Antonsen

High power vacuum electronic amplifiers are candidate sources for high frequency digital communication systems. The manner in which these amplifiers are used will be affected by the nonlinear properties of the amplification process. For conventional modulation schemes non-linearity causes distortion. Alternatively, it has been suggested that modulation of the carrier signal can be achieved by allowing the source to oscillate chaotically but be controlled small perturbations of system parameters. The proposed student research project would entail developing models to characterize the non-linearity and time domain properties of a high power chaotic traveling wave amplifier.

14. Ultra-Violet/Blue/Infra-Red On-Chip Light Emitters and Optical Interconnects
Prof. A. Iliadis

In this project emphasis is placed in developing light emitting devices from wide band gap materials such as GaN, SiC and ZnO on silicon chips for surface and in-plane emission. Silicon-on-insulator (SOI) will be used to provide optical interconnection for in-plane emission, by optical guiding and switching of the optical signals. The light-emitting devices and on-chip optical guides and switches will be modeled and optimized using design software, critical points of interaction for efficient propagation will be identified, and an optimized design will be proposed.

The light emitters will be deposited by Pulsed-Laser-Deposition (PLD), and the emitters and optical guides will be processed and fabricated using standard hotolithographic techniques and processing. Students will have the opportunity to develop an understanding of light generation and propagation through multi-layered structures, the properties of these new and exciting light emitting semiconducting materials, improve their skills in processing and fabrication of relaxed geometry devices, and interaction with industry interested in optical communication and computer systems.

15. Viability of Pre-Fetching in a DBS Scenario
Prof. John Baras

Web pre-fetching has been proposed as a means of enhancing the performance of web browsers, by reducing the end user perceived latency. The pre-fetch technique try to load the local cache of a web browser with possible future pages that might be requested. In this way the pre-fetch technique reduces the latency in a web access to zero, as if the requested document was already in the local cache of the web browser, it will be displayed almost instantaneously. Pre-fetching can be a complicated issue as performance of various prediction techniques is not very well studied.

In this project, we will study the viability of using a simple pre-fetch technique on the actual DirecPC system available in the Hybrid Networks Laboratory. This project will require the students to understand the working of the existing DirecPC system, and then design and implement a software demonstration of a simple pre-fetch technique working on top of the existing system.

Preferred Skills:
---Familiarity with programming concepts C/C++
--Familiarity with network programming, TCP/UDP socket programming