Project Description

1. Introduction

This proposal outlines a plan by Worcester Polytechnic Institute (WPI), Worcester, MA, to use the Abilene Network to establish a high performance Internet connection in support of meritorious research programs and distance learning initiatives which require such connectivity. Applications include spacecraft electric and chemical propulsion, massively parallel processing, Web performance improvement through caching, data mining, surface metrology, ultrasound backscatter, distributed computing, cryptography and wireless LANs, vehicle crash worthiness, and fire protection engineering through distance learning. WPI is a leader in distance learning technology already offering many courses on the Internet both in its graduate programs and in its Division of Continuing Studies. Both research and educational delivery rely on Internet connection and will be enhanced by the Internet2 connection. In addition, many new opportunities will become feasible.

This project will be installed, operated and managed by WPI's College Computer Center. Already a leader with exceptional campus connectivity, multi-media laboratories, and Web-based academic tools, WPI will be able to build on these strengths and become an important collaborator for work necessitating higher speed connections.

WPI is ranked as one of the top U.S. National Universities in the country (U.S. News ranked WPI 48th in 1997 and 51st in 98). It is also one of a very small number of elite technological universities. WPI prides itself in providing outstanding technology in support of research and educational programs. Yahoo! Internet Life Magazine ranked WPI 10th in its annual "America's Most Wired Colleges" survey in May 1998. Distributed via e-mail to over 400 colleges and universities, the survey asked schools what they offered their students in the way of campus-wide networks, e-mail accounts, Web use, and computer lab facilities.

2. Overview

WPI is a privately funded doctoral university that emphasizes technological education in science, engineering and management. Well-known for its excellence in project-based education, the university enrolls 2,700 undergraduate and 1,000 graduate students. A number of research centers, including the Metals Processing Institute, the Center for Wireless Information Technology, and the New England Center for Mixed Signal Processing, reflect strong corporate support and numerous industrial alliances. Strong affiliations with other research organizations, including the University of Massachusetts Medical School, the Central Massachusetts Magnetic Imaging Center, Lawrence Livermore Laboratory, and several NASA Centers provide faculty and graduate students with excellent resources beyond the campus. Located near the high technology beltways west of Boston, WPI enjoys proximity to computer, telecommunications, biotechnology and biomedical companies.

3. Proposed Applications

3.1. Spacecraft Electric and Chemical Propulsion

Dr. Nikos A. Gatsonis, Mechanical Engineering and Director of Space Research at the Computational Gas and Plasma Dynamics Laboratory (CGPL) has developed a strong research program relying on close collaboration with colleagues at other universities and within NASA.

The overall mission of CGPL is to:

Support has been provided by NASA Lewis Research Center and the Jet Propulsion Laboratory; Applied Physics Laboratory for various Ballistic Missile Defense Organization and DOE Space Programs and Missions; NASA's Microgravity Program and the Massachusetts Space Grant Consortium. CGPL involves graduate students, as well as undergraduates completing their Major Qualifying Project (a senior thesis degree requirement at WPI equivalent to three courses). Therefore, the Program has strong educational and research components.

Modeling of Small Gas Thrusters and Comparisons with Space Flight Data. The first part of this study involves 3-D Navier-Stokes and Monte Carlo simulations and requires our interaction with NASA Johnson Space Center. We provide to them simulation inputs from Navier-Stokes solvers and researchers at NASA JSC generate grids for a 3-D Monte Carlo code. Subsequently, we run these Monte Carlo simulations on a multi-processor at Brown University. Increased bandwidth will improve all phases of this interaction and will allow real-time access/visualization of simulation results. The second part of this study involves data analysis and requires access to space flight data and data-analysis programs at Johns Hopkins University Applied Physics Laboratory's computers. Interent2 will clearly benefit this part of our collaboration.

Modeling of Pulsed Plasma Thruster Plumes. The project involves the development of an advanced hybrid code for modeling PPT plumes. Increased bandwidth will allow effective interaction with NASA LeRC as well as research at Ohio State University that is funded under the same NASA program.

Experimental Investigation of Electric Propulsion Thruster Plumes. All CGPL's experiments are conducted at NASA LeRC and JPL facilities and involve graduate and undergraduate students. This integration of undergraduate education and research is utilizing the unique character of WPI's project-based education. Increased bandwidth via Internet2 will allow a more effective communication with students throughout their period of stay at these NASA centers. Transfer of data, remote access of CGPL's computers to perform data analysis, efficient iterations on experimental designs and fast visualization are some potential benefits of Internet2.

Modeling of Plasma Jets in Support of the APEX. APEX is a large space mission with John Hopkins Applied Physics Laboratory (APL), Univ. of New Hampshire, Univ. of Alaska, NASA Goddard Space Flight Center, IGD Russi and WPI. Our role in this mission is to develop a plasma jet model and perform data comparisons. The project will require access of APEX flight data and data-processing programs at APL's computers. In addition, during modeling it is necessary for our APL co-workers to visualize simulation results and provide feedback. It is expected that Internet2 will greatly increase the outcome of this collaboration.

3.2 Massively Parallel Computing

Dr. Homer Walker, Mathematical Sciences Department at WPI is developing an integrated problem-solving environment for massively parallel computers that will carry out full-physics simulation of fire scenarios involving complex reacting flows coupled with structures. Particular target machines are the new massively parallel computers now being brought on line at the national laboratories. Developing scalable numerical algorithms and graphics procedures that will effectively exploit the hundreds of processors on these platforms will be tremendously challenging and will offer extraordinary opportunities for algorithmic research. Work at WPI will focus on development of scalable numerical algorithms and will directly involve collaborators primarily at the University of Utah and Lawrence Livermore National Laboratories and to a lesser extent at Argonne, Sandia, and Los Alamos National Laboratories. Work is now in the initial stages and has not required extraordinary Internet connectivity or usage so far. However, as development progresses, within the next year, we will be making remote runs from WPI on massively parallel machines at the University of Utah and the national laboratories to test the code and ultimately to carry out full-scale simulations. These will require extremely high data-transfer rates for real-time graphical displays of reacting flows and dynamic structural behavior, including fracture and other failure modes. This research will be sharply constrained without much greater bandwidth and lower latency than are offered by current Internet connectivity.

This work will leverage at least two other NSF grants to WPI: (1) NSF Grant DMS 9727128 to WPI, "Iterative Methods for Large Scale Nonlinear and Linear Systems," August 1997--July 2000, H. F. Walker, PI/PD. The C-SAFE problem-solving environment will provide a powerful, highly sophisticated experimental vehicle for research supported by this grant. NSF MRI Grant DMS 9870971 to WPI, "Acquisition of a High-Performance Parallel Computer for Mathematical Sciences Applications," September 1998--August 2001. This grant will enable the acquisition of a parallel high-performance computer in the near future. This machine will be used for local code development and medium-scale testing in the C-SAFE work.

3.3 Improved Web Performance over the Internet through Caching

Professor Craig E. Wills and collaborators Mark Claypool, David Finkel, and Robert Kinicki at WPI together with Dr. Bala Krishnamurthy, AT&T Research Labs and Professor John Hine, Victoria University of Wellington, New Zealand are examining the effectiveness of current caching techniques in light of more complete data, and also to investigate the potential of caching if improved techniques were used by Web caches and servers. There have been many studies to better understand characteristics of the World Wide Web with other studies proposing improved caching policies and mechanisms. However, work has not been done to specifically understand how changes in Web resources and the meta information reported by servers affects caching by Web browsers and proxy caches. To address this gap, we have undertaken a study to monitor and better understand the characteristics of resource changes at servers and how these servers report meta data about the resources. In the initial part of our project - characterizing information about Web resources and server responses that is relevant to Web caching--we have studied a set of URLs at a variety of sites and gathered statistics about the rate and nature of changes correlated with the resource type.

The overall theme of this work is to make more efficient use of the Internet and provide better performance for Web applications through techniques such as improved caching. High performance Internet access is needed for multiple aspects of the project. Our method of probing Web sites directly to collect data is important for the richness of information that is collected, but also costly in the amount of network traffic it generates. We are forced to run our tests overnight because of the considerable amount of time they take to complete. Higher speed access would make our research more efficient in gathering data.

Another need for better Internet access is the periodic need to retrieve Web server and proxy logs. These data sets contain information about requests processed by entities on the Web and are needed to study potential performance enhancements such as improved caching techniques. The logs are large - tens of megabytes and multiple of these logs need to be retrieved.

A final need for better Internet access is for a related project. As part of our work we have built an extensive digital library Web bibliography for literature published on the topic of Web infrastructure and performance. This project both requires tracking journal/conference sites on the Internet, but is also expected to create much network traffic to WPI as other researchers retrieve information from the site contents.

3.4 Data Mining

Drs. Matthew Ward and Elke Rundensteiner are supported under NSF Grant IIS-9732897 to study novel methods for visually exploring and mining very large multivariate data sets (this work is also supported under NSF Research Instrumentation Grant EIA-9729878 with an additional investigator, Dr. Isabel Cruz). The project involves extending current multivariate visualization techniques to accommodate data sets with millions and possibly hundreds of millions of records. This will, in addition, require the design of new interactive tools to assist the exploration of the data as well as data management tools to facilitate the storage of the data and provide rapid access to selected subsets. We are extending an existing visualization package, XmdvTool, which has been under development at WPI for the past 5 years and has been in the public domain since 1994. XmdvTool has been applied to an extensive variety of application areas, and is installed in more than 50 university and industrial facilities.

The basic strategy being followed is to configure the data in a hierarchical structure, using partitioning and/or clustering algorithms as well as hierarchies explicit in the data semantics. Aggregation is used to compute representative information at non-terminal nodes of the hierarchy, and users can explore the data through direct manipulation on either the data display or a visual representation of the hierarchical structure. One of the key issues is attaining rapid access to the data subset being displayed. While small delays would be tolerable, much of the effectiveness of the exploration process will rely on rapid screen updates. To address this, we are examining techniques for caching and prefetching designs based on typical data exploration patterns observed in users.

Once we have achieved acceptable performance on a single computer, we will be extending the system to handle distributed data sources. We are in the initial stages of establishing interactions with researchers and scientists at NASA Goddard Space Flight Center, Woods Hole Oceanographic Institute, and Lucent Technologies to obtain access to some of their large repositories of multivariate data. These and other holders of extensive data archives will provide us a reasonable testbed for studying the applicability of our research to a variety of domains, as well as uncovering bottlenecks in the increasingly important task of distributed visualization for exploring large and complex data sets.

In terms of the demands on the communication network, we hope to avoid transferring entire data sets, which can exceed tens of gigabytes in size. Rather, we will concentrate on the development of database schema to facilitate rapid access to subsets of modest size (on the order of one to tens of megabytes), allowing users to browse the data subspace retrieved and stored in a local cache. Thus the key will be to optimize the speed of data selection and cache update to try and minimize the effect of latency on the user. Clearly, a delay of several seconds for out-of-cache retrievals would be acceptable, but longer delays would negatively influence the exploration process. Existing communication speeds are at least an order of magnitude too slow to allow the effective visual exploration of distributed data sets of the magnitude we envision. Our hope is that the bandwidth and quality of service provided by Internet2 will allow us to create a visualization environment suitable for researchers, scientists, and engineers to effortlessly investigate large, complex data repositories.

3.5 Surface Metrology

Dr. Christopher Brown, in the Mechanical Engineering Department, Manufacturing Engineering Program is involved in the following projects collaborating with the Swiss Federal Laboratory for Materials and Testing in Thun (EMPA-Thun), using, almost entirely, the Internet . We have exchanged large data sets (measured surfaces) and sophisticated software for analysis (area-scale fractal analysis). After four years of work we have experimental support for a new theory (Dr. Christopher Brown's) linking substrate roughness with adhesion, and a new method for determining the scale of interaction (Dr. Stephan Siegmann's).

A food science group at the Catholic University of Santigo wanted to know how to determine surface areas of foods, and found us on the Web last spring. We made measurements and an analysis of potato chips and sent them large data sets via the Internet. The project proved successful and they sent a Ph.D. student to WPI last fall to work further on the collected data. We are still exchanging software and data.

Collaborating with the Chalk River Atomic Energy Lab in Canada we have had a similar exchange of data and software, which has resulted in a paper showing congruence between electro-chemical measurements on zirconia and area-scale fractal analysis of measured surfaces.

In another research initiative we are using special scale-based fractal analyses to characterize surface textures. The characterizations are used to find correlations between the texture and surface creation conditions (e.g., fracture, wear or manufacturing processes) or surface behavior (e.g., adhesion, friction, reflectivity, and chemical reactivity). The surface textures can be measured by a variety of means, including atomic force microscopy, confocal microscopy and scanning laser microscopy. The measured textures generally contain hundreds of thousands of elevations and can be several megabytes in size. Hundreds of measured textures may be used in a project. At WPI, we have the most sophisticated software for performing these analyses, and the most experience using it. Other laboratories often measure the surfaces, and their behavior, then send us the measurements for analysis. The other laboratories are in Switzerland, Canada, Chile, and in the US. Currently we are using Zip disks and CDs to exchange large amounts of information, and FTP or e-mail to exchange smaller amounts. A high performance connection would expand the kinds of projects we could collaborate on and allow us to use our expertise and facilities to full advantage.

3.6 Non-Invasive Imaging for Stroke Research and Atherosclerosis

Dr. Christopher Sotak, Professor of Biomedical Engineering and Director of the Magnetic Resonance Imaging and Spectroscopy Laboratory, conducts much of the most significant stroke research. He is at the forefront of developing new imaging techniques and image processing algorithms. The Laboratory is internationally recognized for the development of MRI/MRS methods for the evaluation of therapeutic interventions in cancer and stroke. The Laboratory is funded by numerous agencies (The Whitaker Foundation, the American Heart Association, the National Institutes of Health and the National Science Foundation) and industrial firms (General Electric, duPont, Schlumberger-Doll Research, HemaGen, Cambridge Neuroscience, Bio-Imaging Technologies, Medical Advances, and Astra Pharmaceuticals). Image transmission is greatly limited by the capabilities of the commercial Internet inhibiting both diagnoses (for patient imaging) and research collaborations (for stroke research). Internet2 accessibility would be a quantum improvement for researchers at WPI, and the University of Massachusetts Medical School (UMMS) (the principal on-site research organizations), and their collaborators across the world.

Dr. Peder C. Pedersen in the Electrical and Computer Engineering Department conducts research into ultrasonic imaging systems and collaborates with Dr. Henri Cuenoud, Cardiovascular Medicine and Drs. Bruce Cutler and Michael J. Rohrer in Vascular Surgery at the University of Massachusetts Medical Center. Two project engineers, Mr. Ron Gatzke and Mr. Ted Fazioli at Hewlett Packard, Andover, MA are also collaborators. The work will soon reach the clinical stage where massive amounts of data will be transmitted between all three organizations.

The goal of the research is the development of a non-invasive ultrasound-based technique for determining the absolute back- and angle-scatter characteristics of carotid plaque. The main objective is to demonstrate improved in vivo classification of carotid artery atherosclerosis relative to what is achievable with current non-invasive diagnostic methods. The proposed technique may lead to better criteria for selecting the patients who will benefit the most from a carotid endarterectomy, thus lowering the incidence rate of strokes. Improved atherosclerotic plaque characterization will be sought by determining the absolute ultrasound backscatter level, including angle dependence, from regions within the atherosclerotic lesion and from the interface between blood and the atherosclerotic lesion, by using the absolute backscatter level of blood as a reference.

The research is carried out in collaboration with the University of Massachusetts Medical Center (UMMC) and the Medical Imaging Division of Hewlett Packard in Andover, MA. The research is currently near the end of the proof-of-concept phase. Subsequent phases will be more clinically oriented and will include in vitro measurement of backscatter signatures from excised arteries, containing atherosclerosis and backscatter signature measurements on excised plaque specimens from endarterectomy procedures, with comparison to histology and pre-operative ultrasound duplex scanning.

In the clinical phases, there will be a need for rapid transfer of moderately large amount of clinical data (100s of MB), in the form of medical ultrasound images and ultrasound RF data, from UMMC to WPI for processing at the Ultrasound Lab. Similarly, we would benefit from the availability of high speed data transfer from UMMC or WPI to HP's facilities in Andover, MA, for sharing of raw data.

3.7 Distributed Computing, Cryptography and Wireless LANs

Researchers in the Department of Electrical and Computer Engineering (ECE) are involved in three areas of investigation that directly impact the future development of applications that will drive the need for high bandwidth and high QoS services for Internet communications. The following summarizes key efforts of three members of the ECE department in distributed computing, cryptography and wireless LANs. The combined result of these thrusts will be mechanisms that enable new kinds of applications that will deliver secure and reliable multimedia and fiduciary data streams to fixed and mobile users of immense long-term economic importance. Availability of high speed and high QoS services will allow each of these researchers to expand their efforts and introduce larger scale and more collaborative tests of their new technologies.

Professor David Cyganski heads the Convergent Technology Center, which engages in the exploration and application of the converging technologies of computing, communications and cognition. Research is focused on the development of new algorithms, protocols for fault tolerant Internet working and low latency, real-time multimedia transport, and exploitation of advances in distributed processing, middleware and network communication technologies. The goal of the CTC is to rapidly move emergent technologies into commercial, medical, and defense-related applications for its sponsors.

Past and current projects explored the performance and optimization of network protocols for real-time data and multimedia communication, distributed object services based on CORBA, ActiveX and Java, fault tolerant multi-tier systems and networks, and realization of distributed, multi-platform, virtual-reality based visualization using real-time and persistent object information fusion with geographic information system resources.

A recently undertaken thrust involves exploitation of Internet worked computers to support the creation of distributed real-time sensor/processor systems for industrial machine vision and control and military automatic target recognition.

Professor Christof Paar heads the Cryptography and Information Security (CRIS) Group. The field of cryptography and data security is shifting from an area with mainly military applications to an area with enormous potential for numerous applications in the commercial sector. The research efforts in this group have two main foci: system integration of cryptography in modern applications, such as wireless or ATM networks; development of faster implementations of new cryptographic algorithms.

Most public-key crypto algorithms are based on arithmetic with very long numbers (150-1000 bits). One goal of the CRIS group is the development of a new generation of arithmetic architectures which are faster than previous ones yet area efficient. This includes development and implementation ASICs as well as FPGA-based architectures. Another emphasis is system integration of crypto algorithms, protocol development, and secure network management.

Professor Kaveh Pahlavan heads the Center for Wireless Information Network Studies (CWINS). The primary objectives of the Center include: performing basic research in technical aspects of wireless information networks; participation in the development of experimental test beds and performance monitoring tools; developing experimental sites and performance evaluation facilities for comparative studies of alternative wireless systems; organizing workshops, symposia and short courses to educate engineers about wireless networks and create an atmosphere conducive to information exchange among the participating corporations; coordinating basic research and work on regulatory issues in the wireless information network industry and the portable mobile industry; providing on-site training to engineers in the industry.

The Wireless LANs Technology Group operates within CWINS, concentrating on indoor wireless LANs and building-to-building bridging systems. While initial technological interest is focused on infrared, the ISM bands, and ETSI 300.328 bandwidth for Europe (and similar bands in the rest of the world), unlicensed PCS and suitable licensed bandwidth is also considered within the scope of the WLTG. An experimental testbed for performance monitoring of the wireless LANs and wireless bridges has been developed in the form of a modern wireless classroom with a campus-wide wireless access. This project provides an example of a successful application and increases the awareness of the benefits of wireless technology.

3.8 Vehicle Crashworthiness Research

Dr. Malcolm Ray and other researchers in Civil and Environmental Engineering are involved in vehicle crashworthiness research projects that use the explicit nonlinear finite element program DYNA3D and LS-DYNA to numerically simulate collision events.

In particular, Dr. Ray is leading a Federal Highway Administration effort titled "Side Impact: Finalizing the Test and Evaluation Procedures." This project is actually a collaboration between the Federal Highway Administration and the National Highway Traffic Safety Administration. The purpose of the project is to use finite element simulation to obtain a better understanding of occupant injury mechanics in real-world side impact collisions with roadside objects. Such collisions fatally injure approximately 1,500 annually people and are responsible for at least three billion dollars of societal loss each year. The project is exploring the mechanics of human injuries in side impact collisions by using large complicated finite element models of anthropometric test devices (e.g., test dummies), vehicles, and traffic barriers. These models are generally on the order of 100 to 150,000 finite elements and the simulations are carried out for between 100 and 500 msec using one microsecond time steps. While very detailed and highly accurate, these models are computationally demanding and they generate large quantities of data about the stresses, strains and kinematics of each element.

Dr. Ray is leading another effort co-sponsored by the Federal Highway Administration and the Iowa Department of Transportation to examine and improve the performance of one of the most common barrier systems used in the United States, the strong-post W-beam guardrail. This system has been used in the US for nearly 50 years but recent full-scale crash tests have demonstrated that the system does not perform well with some of today's newer vehicles. Finite element simulation provides a method to explore the performance of the barrier system and possible retrofits to the design without performing as many expensive full-scale crash tests.

Public-domain crashworthiness research projects like these have evolved such that there is extensive collaboration between researchers on developing and modifying models. For example, in the FHWA Side Impact project the project team worked with the National Crash Analysis Center at George Washington University to obtain and refine the vehicle model, with SRI to improve its side impact anthropometric dummy model, and with Livermore Software Technology Corporation to add and refine features of the analysis code needed for the simulations. These other organizations were involved with the original development of the models or the analysis code and collaborating with them to add enhancements and improvements was the most cost-effective manner of performing the research. There is an informal group of universities and government laboratories involved in collaborating on the use of finite element methods in crashworthiness research including WPI, the National Crash Analysis Center at George Washington University, the Federal Highway Administration's Turner-Fairbank Highway Research Center, NHTSA's Volpe Center, Livermore Software Technology Corporation, Lawrence Livermore National Laboratory, the University of Iowa, Texas Transportation Institute at Texas A&M University, the University of Nebraska's Midwest Roadside Research Facility, Florida State University, the University of Cincinnati, and others.

Unfortunately, the size of the data files generated by these models and the computational resources required to perform an analysis are considerable and therefore inhibit effective collaboration. A higher performance Internet connection would help reduce several specific obstacles to collaboration: (1) data transfer time, (2) interactive collaboration, and (3) access to off-site computational resources. First, a typical problem generates over one gigabyte of binary data and it is not unusual to generate two and three gigabytes in more complicated runs. When collaborating with the original developer of a model to improve the model performance or add features, it is essential that the results of analysis runs can be shared such that all collaborators can explore the results. Sharing some analysis results is not feasible at this time because ftp transfer of data may take more than 24 hours. Improved network performance would reduce the time required to exchange data between collaborators and thereby enhance the quantity and quality of the collaborations. Second, animations and post-processed results can be shared across the network such that collaborators can discuss and jointly explore the analysis results. Such joint viewing has been used on occasion; however, the demands made on the network have resulted in poor quality interactions. Third, WPI does not have sufficient high-performance computational resources on its own to perform many of these analysis runs. Supercomputer and parallel processing resources at other organizations can be used to perform analyses, but since the resulting output are large it is difficult to either transfer the data back to WPI or remotely view the results over the network. In short, difficulty in moving large binary data files is a significant inhibitor to technical progress and effective collaboration in the crashworthiness research area. Higher speed, higher quality network connections would have a significant impact on WPI's ability to perform this type of research and work more effectively with our collaborators.

3.9 Fire Protection Engineering Through Distributed Education and Research

For decades, WPI has utilized a variety of modern communication and instructional technologies to successfully provide high quality distance learning opportunities through its Advanced Distance Learning Network (ADLN). In 1979, WPI became the first school in the country to offer for-credit graduate management education via distance learning technology. This bold endeavor was followed up by WPI's Center for Firesafety Studies in 1993 when it became the first in the world to offer its unique educational program in fire protection engineering (FPE) to practitioners at widely scattered national and international locations. Funding for our activities comes from an array of sources including the Davis Educational Foundation, NSF, and NIST.

Professor David Lucht, Director of the WPI Center for Firesafety Studies is known worldwide for his leadership role in fire research technology transfer and education. In 1974, the Fire Prevention and Control Act prompted the creation of the United States Fire Administration. Professor Lucht, was appointed by President Ford (and later re-appointed by the Carter Administration) to serve as the deputy administrator and oversee the startup of this agency as well as the National Fire Academy and the National Fire Data Center. As stated by the President's Commission in its report to Congress entitled America's Burning, "Appallingly, the most richest and most technologically advanced nation in the world leads all the major industrialized countries in per capita deaths and property loss from fire." Professor Lucht and his colleagues in the Fire Protection Engineering Program have exceptional capability in creating new engineering tools, models and calculation methods; researching reforming building code regulations to allow for innovative new processes; and training engineering specialists in the use of new technology. Previous NSF support has enabled WPI to play a unique role in all of these areas. However, more support and collaboration is needed so that universities can join forces to meet global needs and support regulatory reform. WPI is strategically positioned to attack these critical areas. An Internet2/Abilene high performance connection to WPI would allow the Center for Firesafety Studies to expand its offering of these resources and accelerate the flow of research results to researchers and practitioners and all over the world.

WPI will take the lead to create a global learning and research collaboration network to share data with other researchers, work collaboratively with other university and industry partners to uncover new knowledge about fire behavior and fire protection methods and deliver graduate education programs in fire protection engineering. The new Fire Science Laboratory at WPI offers a state-of-the-art fire research and testing capabilities that could be accessed electronically for remote experimentation and data transfer if the bandwidth and QoS of Internet2 was extended to WPI. Unique resources such as WPI's Cone Calorimeter, LIFT Apparatus and Smoke Density Chamber along with the new lab equipment purchased with support from NSF (Phase Doppler Particle Analyzer and Infrared Imaging System) could be made available through electronic means to fire researchers, practitioners and graduate students all over the world. WPI could also offer enhanced on-line access to state-of-the-art fire modeling software and access to more than 11 million records of fire incidents in the U.S. using the National Fire Incident Reporting System (NFIRS) database. Real-time transfer and analysis of testing data produced by WPI fire modeling software or laboratory equipment, located in the Fire Science Lab, could be made accessible to a worldwide community if a reliable high-speed network connection to WPI was established.

The Center for Firesafety Studies and the Department of Management (Management Department Chair and Professor McRae Banks, and Professor Wenhong Luo) have also begun to forge strategic partnerships with other universities to co-develop instructional programs and cooperate on other areas such as faculty and student exchanges, research, symposia, and theses and undergraduate projects. Agreements with Chulalongkorn University, Thailand (Dr. Kitti Intaranont, Faculty of Engineering)and the Federal University of Pernambuco, Brazil (Professor Dayse)currently exist and graduate level fire protection engineering courses and programs for students and practicing professional in these countries are under development. Additionally, the Department of Management is involved in discussions with universities in Venezuela, Columbia, Ecuador, Peru and Argentina for collaboration on distance-based graduate management programs and courses. WPI's Advanced Distance Learning Network (ADLN) infrastructure, managed by Ms. Pennie Turgeon, is currently used to deliver asynchronous and synchronous classroom lectures and supplemental resources to these and other remote participants, but with limitations. Videotape, while it can store a large amount of information, is too passive for today's learning environment and delivery to off-site participants introduces sync issues that are magnified by courier limitations. Interactive videoconferencing over ISDN lines does not provide satisfactory video and graphics resolution when used at lower bandwidths; at higher bandwidths, it's just not cost effective. It also does not provide an adequate assortment of interactive multimedia options. Internet -based audio/video technologies are often the essential tools for "broadcasting" lectures and rendering multimedia materials on demand. For distance learning activities, video conferencing and streaming video products are the most prominent existing Web-based audio/video technologies, but due to limited Internet bandwidth and the lack of traffic control, the quality of audio/video delivered via the Internet is far from satisfactory. The quality of audio sound and video image is poor and unpredictable. In addition, most existing interactive video products provide only limited capability in terms of supporting interactions among students and the instructor. The use of the Internet to combine the best of all these communication tools holds promise, but today's networking, desktop computing and multimedia technologies, though improving rapidly, are not quite ready to support wide scale "anytime, anyplace" learning and research. Thus, an important research question in distance education is how we can improve the quality of the virtual learning environment if the performance of the connection is improved.

WPI has made the most out of the available technologies, but the widespread deployment of the Internet in its current form has remained at arm's length due to limitations associated with bandwidth and systems integration. Internet2 QoS technologies and multicasting, coupled with new applications for instructional management, virtual laboratories and collaborative consortia, tele-immersion, and digital libraries promises to change this situation rapidly and dramatically. Assessment of new applications and their effectiveness in furthering learning will be very important. WPI has done studies to assess the outcomes of learning experiments using advanced technologies (e.g., Developing Retargetable Web Software for Course Delivery). Some of the work has been funded by the Davis Educational Foundation and directed by Professor Judith Miller, Director of the Center for Educational Development.

Our research project aims to develop and test Web-based interactive video software that supports various approaches of real-time interactions based on streaming technology and the Internet2 standards. The resultant software should provide better experience of real-time interactions for both the students and the instructor. More importantly, the software will give the instructor more choices as to how he wants students to interact in a virtual environment. For instance, the software may allow the instructor to create virtual groups during a lecture so that a student can only see and hear those in his group. When group discussion is finished, the instructor can switch back to the classroom mode. The testing of the software will be conducted in collaboration with faculty members (e.g., Dr. Hao Lou, College of Business) at Ohio University.

To successfully develop and test the kind of software proposed in our project, we need both the bandwidth and QoS guarantee provided by an Internet2 connection. With the QoS guarantee, we can have better control of the quality of the multimedia presentations and also ensure that all students will receive the same information. While more bandwidth is needed for better video quality, it is also required if we are going to implement various approaches of real-time interactions via the Internet. Many of the real-time interactions required for the implementation of project-based cooperative learning techniques will easily double or triple the bandwidth requirements. For example, one design may call for the instructor to have control of whether students can see the instructor only, the student who is asking a question, and/or the entire class. To implement this design, we may have to first collect live video signals from every student and then stream video back to them. This would require significantly more bandwidth. Taken together, high performance Internet connectivity is a necessary condition for the development and testing of web-based, real-time, interactive application software.

Meritorious Research Application Area Data Set Size
(Megabits)
Bandwidth Need
(Mbit/s)
LatencyNeeds
(ms)
Frequency of Use Duration of Use
3.1 Spacecraft Electric and Chemical Propulsion 2000+ Real-time OC-3 ATM Minimal 3-4/Wk Hours
3.2 Massively Parallel Computing 1000+ 100 Under 300 ms Weekly 8 Hours
3.3 Improved Web Performance over the Internet Through Caching 100's 100 Minimal Daily 7 Hours
3.4 Data Mining Research 50's 100 Minimal Daily 5 Hours
3.5 Surface Metrology 10's 100 Minimal Daily 10 Hours
3.6 Non-invasive Imaging for Stroke Research and Atherosclerosis 5's 10 Minimal Daily 10 Hours
3.7 Distributed Computing, Cryptography, and Wireless LANs Not Applicable Real-time OC-3 ATM Under 300 ms 3-4/Wk Hours
3.8 Vehicle Crashworthiness Research 2000+ 100 Minimal 3-4/Wk Hours
3.9 Fire Protection Engineering Through Distributed Education and Research Not Applicable Real-time100 Mbit/s Under 300 ms Daily 10 Hours

3.10 Other Meritorious Applications

Many other research programs, while not presently utilizing Internet connectivity, nevertheless, will mature into significant users or would use I2 where the commercial Internet is unsuitable. Brief descriptions of these applications follow:

The Metal Processing Institute, under the direction of Dr. Diran Apelian, Howmet Professor of Mechanical Engineering, is the largest industry alliance in North America. Three consortia constitute the MPI: the Aluminum Casting Research Laboratory, the Powder Metallurgy Research Center, and the Semisolid Materials Processing Center. Near net shape manufacturing is a major thrust of research efforts resulting in transmission of large CAD and data files through the Internet. The interactive mode of communication enabled by the Internet has transformed the alliance from a good concept to an effective and highly relevant institute serving the needs of the industrial sector by advancing the knowledge-base. Internet2 and high performance connection will greatly facilitate a national laboratory and other university collaborations.

ODEToolkit Project. The ODEToolkit project has been an on-going collaborative research project between Harvey Mudd College and WPI since January 1997. Professor Joseph Fehribach, Associate Professor in Mathematical Sciences at WPI is the lead researcher on this project. The research has been sponsored by Digital Equipment Corporation (now COMPAQ/Digital), and has been carried out primarily by teams of students at WPI and Harvey Mudd College. The goal of our collaboration up to this time has been to establish an ordinary differential equations (ODE) solver on the web, giving convenient access to anyone with a modern web browser to state-of-the-art methods for solving initial value problems for systems of nonlinear ODE. What is new here is the use of the Web and the Internet to conveniently bring these methods to large audiences at no charge. The solvers run on servers at WPI and Harvey Mudd, while clients running locally on the users' computers handle data manipulation and the graphical user interface (GUI). Only the minimum amount of data is transferred between the client and the server. Currently beta versions of ODEToolkit are available at http://ODEToolkit.wpi.edu and http://ODEToolkit.hmc.edu.

The ODEToolkit project can benefit from the greater bandwidth of the Internet2 in several ways: The Internet2 will enable more-effective day-to-day collaboration between Harvey Mudd and WPI on future design and development. While only a small amount of data is transferred between the client and the server when ODEToolkit is in use, our teams have often been limited in our ability to transfer large amounts of graphics, so that they can simultaneously view and manipulate the same images. This limitation has been a major problem which we have overcome only through a great deal of patience and understanding. In addition, while any one user would only require a very small amount of bandwidth to use ODEToolkit, it is our hope that our servers become widely used both by students beginning to study ODE, and by researchers who come across a particular ODE system in their work and wish to have a quick, easy-to-use tool for investigating its solution. Through grants from Digital and other vendors, we can obtain servers powerful enough to handle tens or even hundreds of thousands of simultaneous users. Limitations on our network bandwidth, however, may be much more difficult to overcome. Internet2 connectivity would greatly help in this area, at least for all users at other Internet2 institutions. Finally wider bandwidth would allow a number of future developments to ODEToolkit. For example, given sufficient bandwidth, one could design a feature in ODEToolkit which would enable users to handle real-time telemetry from a physical system, and simultaneously study this system and an ODE model being solved by one of our servers.

Global Project Centers. WPI operates a large network of project centers in many countries in order to support its emphasis on global education. Half of all undergraduate students spend one term in residence at the centers. Project advising, at a distance, is becoming very important in order to ensure that students work with advisors who are expert in the project subject area. For the most part, this means, advisors are on campus while students are in residence at the project centers. Use of the Internet for project advising purposes including e-mail, voice, video and graphics communication, is growing rapidly. The Internet offers substantial limitation on these educational practices, which would be overcome by Internet2.

Digital Library. WPI's Gordon Library is one of our greatest intellectual assets as well as a critical element of our campus culture, just as our high-performance networking infrastructure is now critical to nearly everything that we do. The library provides access to a breadth of data, information and knowledge for science, engineering, management, social sciences and the arts. For example, the library homepage on the WPI Website provides access to hundreds of electronic journals, remote and local databases, the library catalog, as well as electronic reference, interlibrary loan and acquisitions services. Our extremely competent staff work closely with faculty, research staff, and students to facilitate their research and knowledge of our electronic infrastructure.

We are members of critical consortia: the Northeast Research Library Consortium (NERL), the Worcester Area Cooperating Libraries (WACL) and the Central Massachusetts Regional Library System (CMRLS). We have worked diligently with these groups to minimize expenditures for overlapping holdings and now various libraries share multiple electronic resources over the Internet. The Internet2 project will allow more extensive sharing, much faster access, and stimulate projects for digitizing holdings such as our unique collection of Dickens materials. WPI is also planning to become a contributing member of the Networked Digital Library of Theses and Dissertations and to help create standards and processes for our electronic submissions. We hope to participate in Internet2 digital library initiatives as well. Implementation of our plan to become the Worcester Internet2 gigaPoP puts WPI in a leadership position to bring the benefits of the future web of digital libraries to the desktop.

4. Contribution to Emerging National and Global High-performance Network Infrastructure

The Connections to the Internet Program has come at a most fortuitous time for WPI. Our President, faculty, staff and university management have debated our strategic direction for over two years. The net result is a bold 10-year Strategic Plan for the university. It paints a bright vision of the future and defines five specific mission areas--each with concrete goals and actions required to accomplish our strategic goals.

There is no doubt that the Internet2 program and a high performance connection are critical elements in the achievement of all five of our strategic missions and goals. For example, we seek to improve WPI's status as a national university and move from a Doctoral II level institution to a Doctoral I institution. This will require further emphasis in our graduate program, substantial increases in our volume of research, and more collaboration among existing and future Internet colleagues. The WPI Internet2 and high performance connection will drive new research initiatives and invigorate existing ones as we begin to share our physical resources (e.g., FPE laboratory, NMR lab equipment, digital library holdings, etc.) and receive access to remote resources such as supercomputers and other laboratory devices. The Internet2 and Abilene network will also act as a test bed for new internetworking technologies being developed by our ECE researchers (Section C); results will also be shared among the community, and possibly incorporated into new standards, products and services.

Another strategic mission is to further establish WPI as a leader in global technological education. For decades, WPI has operated one of the most extensive undergraduate global studies and advising programs. We operate project centers distributed across the USA, Europe, and Asia with an extensive network of collaborators in the global community of university researchers, students and corporations. Internet2 connectivity and a high performance network will greatly accelerate the synergy gained from these multi-disciplinary and multi-cultural relationships. Further USA connectivity between peers such as Internet2 and vBNS is being negotiated and is a critical relationship. Internet2 is negotiating connectivity with various Canadian (www.canarie.ca) and European networks (www.nordu.net and www.surfbureau.nl) and there are thoughts about Asia as well. The major benefit will be to share resources, knowledge, ideas, and plans in a more concrete way using new technological tools emerging from the Internet2 applications development work. WPI's contribution will be that of an active player in these activities who builds upon its current strengths, while sharing with our collaborators.

WPI will contribute to the emerging national network infrastructure by establishing a Worcester Internet2 gigaPoP and associated regional network. In collaboration with NEESCom, we plan to build a gigaPoP in the NEESCom Exchange Building, located at 474 Main Street, Worcester MA, which is a hub for 30 route miles (7000 fiber miles) of their fiber ring. As an example of how well-connected this location will be, Qwest is planning a demarcation point here. This is attractive because it is a "free trade zone" and many other providers are expected to co-locate. The building is designed for this purpose and has over 3MW of power and associated technical and physical services (e.g., backup power, HVAC, etc.). As an element of our participation in this project, we will encourage the involvement of the 9 other universities in the Worcester area. All of them are smaller scale than most Internet2 institutions; among them are Holy Cross College, Assumption College, and Clark University. This collaboration would constitute a laboratory for evaluating the deployment and performance of Internet2 applications by institutions of this size. NEESCom will facilitate this activity by providing highly aggressive pricing for their connections to their fiber ring.

WPI's commitment to local, national, and global internetworking development is demonstrated in many ways. We plan to further develop our high performance intranet infrastructure, build an Internet2 gigaPoP and associated regional network, continue to develop advanced internetworking software and techniques for fault tolerance, cryptography and wireless LANS, participate in Internet2 activities, and use the emerging global connectivity to quickly pursue research expansion through collaboration--a critical strategic goal of the university.

5. Network Engineering Process, Participants, and Plan

5.1 Planning group and process

The network planning to incorporate the Internet2 OC-3 ATM connection was a joint effort between WPI Information Technology management, Network Operations Staff, Nortel Networks, Qwest, and New England Electric Service Company (NEESCom). The WPI Network Operations Staff, consisting of Network Manager, Mr. Sean O'Connor, Co-PI and five Network Technicians, along with Mr. Allan Johannesen, Managing Senior Unix Administrator, and Co-PI, Dr. Thomas J. Lynch, Vice President for Information Technology, PI, worked in conjunction with Nortel Network Enterprise System Engineers, Gary Cattarin and Todd Kelleher, to incorporate the Abilene connection within WPI's current network infrastructure.

WPI also worked closely with Paul Keleher from NEESCom and Scott MacCloy from Qwest concerning Abilene connectivity and fiber placement and availability for WPI as well as other educational institutions and corporations in the area.

5.2 Network Engineering and Design

WPI networking staff, partnered with NEESCom and Qwest, has engineered a plan to use the NEESCom fiber infrastructure and the Qwest ATM connectivity to connect to the Abilene network in a cost-effective manner (Figure 1). A Nortel Networks Backbone Concentrator Node (BCN) router will be on the WPI campus. A BCN is a symmetric multiprocessing router with 4 backplane slots. It will be installed in the location of the campus border router, which is currently a Cisco 3600. The campus commodity Internet links will still be connected to this router. This campus router will connect via OC-3 ATM over NEESCom fiber infrastructure to the WPI gigaPoP router to be located at the NEESCom fiber concentration point at the NEESCom Exchange Building, 474 Main Street in Worcester, MA.

This is an ideal location for the gigaPoP, since it is the center of the NEESCom fiber rings which run throughout Worcester, offering easy interconnection of area universities and research institutions to the gigaPoP via that same, cost-effective NEESCom fiber. NEESCom is partnering with WPI in the gigaPoP, so they are offering WPI the free use of its fiber for our interconnection. The gigaPoP router will be a Nortel Networks Backbone Link Node (BLN) router. A BLN is a symmetric multiprocessing router with 13 backplane slots. Symmetric multiprocessing, a feature of both the BCN and BCN, means that as communications cards are added to the backplane, CPU power is also added with each card, so router power is constantly increasing as interconnections are added. This enables connection growth without performance degradation. The gigaPoP connection to Qwest will be made at the NEESCom fiber interconnection point. The link from the WPI campus to the fiber concentration point offers great flexibility to the campus. WPI might decide to apportion part of the ATM link between WPI and the gigaPoP as a feed of commodity Internet to the campus, connecting to Qwest, GTE Internet working, or other area provider, at this fiber concentration location.

To enforce the Internet2 Acceptable Use Policy, WPI will leverage its current ATM infrastructure to provide a separate Internet2 Emulated LAN (ELAN) for Internet2 traffic. WPI will offer OC-3 ATM or 100 Megabit switched Ethernet connections to meritorious Internet2 projects.

6. Local Network Infrastructure

The current WPI Network Infrastructure (Figure 2) consists of an OC-12 Private Network to Network Interface (PNNI) redundant Backbone with a Nortel Networks Centillion C1600, an ATM switch, at the campus fiber concentration point. The Centillion C1600 is connected in an OC-12 ATM ring to three System 5000 BHC chassis located in Fuller Labs. The C1600 switches PNNI OC-3 ATM connections to fourteen residence halls and six academic buildings at which WPI offers 10 or 100 megabit switched and shared connectivity to the desktop. There are about 4,100 nodes on the network.

Two Centillion 100 ATM switches, with dual OC-3 uplinks to the C1600, provide 10 and 100 megabit switched fiber connections to thirteen smaller academic buildings and three residence halls.

The WPI Greek community, consisting of twenty fraternities and sororities is connected to the WPI LAN via 10 megabit wireless links. One WPI administrative building located off the central campus, but within line-of-sight, is also connected with this wireless setup.

WPI's two satellite campuses, located in Waltham and Southboro, are connected via compressed T1's to our Nortel Networks BCN router. The BCN, which will also handle WPI's two commodity Internet connections, is connected to the backbone via an OC-3 uplink. This router, as mentioned above, will be the campus connection point to the gigaPoP.

To regulate and monitor traffic between the academic network, residential buildings, Greek houses, and external campuses, WPI has defined separate ELANs for each area. In addition there is a network management ELAN to provide secure network management. All ELANs are connected via a Nortel Networks redundant Virtual Network Router (VNR). Since ELANs are already part of WPI's network architecture, it will be easy to integrate the Internet2 ELAN, mentioned above.

WPI's current three-year infrastructure upgrade plan is to change the backbone to a quad-OC12 ATM architecture. This will entail moving the C1600 ATM switch from the fiber concentration point to Fuller Labs and replacing it with the Nortel Networks Raptillion ATM switch. The relocation of the C1600 will allow ATM capable WPI servers to be upgraded to OC-3 ATM uplinks. Also included in the plan is to upgrade all academic building uplinks to OC-3 ATM and replace current network equipment with System 5000 BHC or System 5005 BHC chassis. Internal academic and staff network building wiring will be upgraded from coaxial cable to category 5 level 7 enhanced cable. This will allow Abilene/Internet2 access to all current and future meritorious projects.

7. Quality of Service Guarantees and Implementation Methods

The Centillion switching solutions from Bay Networks support transmission of voice, video, and conventional data traffic in campus ATM networks and over both public and private ATM LANs and WANs. Centillion 1000 multi-service switches fully support all ATM QoS classes. Delay-sensitive Constant Bit Rate (CBR) voice and video traffic is assigned higher priority to ensure the required bandwidth and minimal buffering to achieve successful delivery. Variable Bit Rate (VBR-rt and VBR-nrt) traffic has less stringent delivery requirements, and may be buffered in output queues. Unspecified Bit Rate (UBR) and Available Bit Rate (ABR) traffic is not latency-sensitive, and accepts best-effort delivery whenever bandwidth becomes available. The Centillion 100 and System 5000-BHC ATM switching solutions also provide QoS support, providing CBR, VBR-rt, VBR-nrt, and UBR classes of service. Call setup is performed in conjunction with PNNI, which is deployed throughout the Centillion switches, to assure highly available services.

Centillion 1000 multi-service switches provide strict traffic policing services, ensuring that all flows adhere to their service contracts so that service levels across the network are not compromised. Centillion 1000 traffic management capabilities include Explicit Forward Congestion Indicator (EFCI) support, as well as Explicit Rate, No Increase, and Congestion indicators in the resource management cells of ABR traffic.

With the QoS support provided on the Centillion family, the network offers end to end QoS. For example, a PC can signal a CBR call, and the call will be processed at the highest priority with guaranteed bandwidth to the destination. With QoS aware PNNI routing, the traffic contract initiated by the PC is endorsed end to end on every switch, to the destination. The effect is a large scaleable network that honors the traffic contract.

When it is available in late spring / early summer 1999, WPI will deploy revision 4.0 Centillion software, providing MPOA (Multi-Protocol over ATM) Client (MPC) functions within the LAN. The MPOA Server (MPS) function is already available in the BayRS routing code currently running in all campus routers, and need only be activated when required. MPOA will provide traffic-level-sensitive flow-based cut-through route switching (akin to Layer 3 switching), reducing routing latency, increasing IP throughput, and enhancing all other QoS functions through this improved performance.

Tuning parameters within the MPOA functionality will allow increased control over the quality of services provided for meritorious research traffic.

8. Technical Expertise and Coordination with Network Service Providers

8.1 Technical Expertise

WPI has been on the Internet for approximately 10 years. Prior to that, WPI was on CSNet and BITNET, both at only 9600 baud. Our initial provider was NEARnet, a subsidiary of BBN. NEARnet later became BBN Planet, which was acquired by GTE and is now known as GTE Internetworking. For about 8 years, this provider had their Worcester PoP at WPI. Our staff acted as "hands and eyes" for them. For example, we might occasionally place cards in test mode, or configure wiring, as directed by the ISP, saving them traveling from Cambridge to Worcester. We were not involved in programming their routers, however. WPI's initial Internet connection was 56K which we later upgraded to a T1. Two years ago, we searched for a second provider for more bandwidth as well as redundancy. We chose MCI, which has since become Cable & Wireless. In order to bring that second provider on-line, we acquired an ASN for WPI so we could run BGP in the router, as required with more than one provider, and we acquired our own Cisco router and began to manage it.

We have always configured and managed our own servers, e.g., name server, mail server (including pop and imap), ftp server, news server, web server, proxy server, etc. Since our last campus network infrastructure upgrade, about 3 years ago, the campus backbone has been ATM (currently OC-12), based on Bay Networks equipment, so we are highly familiar with ATM as well.

8.2 Coordination with Network Service Providers

We are in the process of negotiating for two more T1 commodity lines and the WPI administration has agreed to support the addition of two more over this summer, if traffic warrants. We have two T1 lines now, so we have a potential of a 300% increase in commodity Internet bandwidth for our campus in the coming months. Since we already have an ASN, control our own router, and are running BGP, we can connect to a new provider at any time, or increase the number of lines to current providers, so we have flexibility in our negotiations. WPI has an outstanding relationship with Bay Networks (Nortel), who have been our prime provider of campus network equipment in recent years, and will be switching from a Cisco router to a Nortel router when we begin our T1 reconfiguration, mentioned above. Nortel has indicated that they will support our effort to connect to Abilene.

9. Service Availability and Continuing Support

9.1 Service Availability

Given the distribution of Internet2 via ELAN over the campus ATM backbone, WPI is enabled to provide access to the Internet2 at any university location. An Internet2 committee led by Dr. Thomas Lynch, PI, will evaluate researchers' applications for access and their requested quality of service. The committee will have access to Internet2 traffic reports from Network Operations.

Qualified projects could connect to the Internet2 via OC-3 ATM, 100 megabit switched Ethernet, or 10 megabit switched Ethernet, as the committee determines. The link would be provided via the System 5000 BHC or System 5005 BHC chassis.

9.2 Continuing Support

WPI's President, Edward Parrish, our Board of Trustees, and other university administration understand that a commitment to networking is essential in a technological university. To demonstrate that, in the last 5 years, the campus infrastructure has evolved from an FDDI ring, to OC-3 ATM, to OC-12 ATM. As proof of that continuing commitment, WPI has planned a campus-wide multi-year upgrade of networking infrastructure, as described in section 6.0.

Relative to Internet2, in partnership with NEESCom, we plan to sign subscribers to our gigaPoP so that continued participation should be economically feasible, so that our researchers can enjoy its benefits beyond the term of this grant.

10. Cost Effectiveness, Thoroughness of Effort, and Proposed Cost Sharing

Please see Section F, Budget Summary and Justification, for a detailed analysis of our proposed cost sharing and demonstration of our commitment to a cost-effective solution. Here we simply summarize.

WPI has proposed an outstanding and extremely cost-effective solution. Our approach has been thorough, and we have maximized our situation by requesting vendor donations of OC-3 fiber and connections, physical space for our gigaPop, an ATM router and associated electronics, and network engineering time. This will save us at least $200,000. We have explored a variety of architectural concepts, discussed options with many vendors, and are seeking to establish ourselves as the Internet2 gigaPoP in Worcester. This will have the effect of significantly lowering our cost and the cost for the many other local universities, colleges, and research institutions as they obtain high performance Internet2 connectivity. We have chosen to partner with New England Electric Service Company (NEESCom) to obtain the best architecture and network performance, at the lowest cost available to us, at this time.

It is proposed that NSF fund a portion of the costs required for an OC-3 connection to the Abilene network in Worcester. Section F contains a detailed, yearly breakdown of the cost sharing and items to be purchased. WPI's co-funding level of $272,500 is greater than that requested of NSF ($260,000). Note that our planned three-year infrastructure upgrade program will devote approximately $2.8M to further enhance the effectiveness of the requested high performance connection.

11. Research Evaluation and Dissemination

WPI will evaluate and disseminate results and information about our research, educational programs, and technologies associated with the use of a high performance connection. We will maintain an active series of web pages over the life of the award. It will describe the overall project, our meritorious research with links to collaborators, our high performance network connection with internal topology and equipment, QoS support and performance measurements, our positive experiences and lessons learned, contact information, ways to collaborate and share resources with or at WPI. Our project evaluation will focus upon results and how the high performance connection affects:

Work in the above areas will be evaluated by a group of faculty and technical staff led by Dr. Thomas Lynch, principal investigator, Vice President for Information Technology, and Professor Judith Miller, Director of our Center for Educational Development and Technology Assessment. Project goals will be established with metrics to measure performance. In addition, to a highly dynamic web site, WPI will employ the following methodologies to further disseminate results and to inform the Internet working community:

Maintained by itweb
Last modified: Jun 07, 2010, 15:22 EDT
[WPI] [IT] [Back] [Top]