INFRASTRUCTURE 93

     We propose to establish an institutional infrastructure at Florida International University (FIU) that will serve as a platform for innovation in computer science and engineering research and education. This infrastructure will be managed by a Center for Advanced Technology and Education (CATE), which will develop a state-of-the-art research facility to integrate critical technology areas supported by the CISE Directorate. CATE will unify projects in software engineering, computer vision, neural networks, artificial intelligence and robotics applications, and computer-aided education under the theme of concurrent processing, an important technology enabler for each area. Since a principal objective is to enhance the educational and research activities of our students, we will revitalize the graduate and undergraduate curricula in the cited disciplines. We will, further, provide our students with innovative and challenging research opportunities in emerging and critical technology fields.

     Our strategy is to take advantage of existing strengths in key areas supported by the CISE Directorate. CATE will encourage research synergy by pooling resources and promoting interdisciplinary projects. The proposed infrastructure will greatly improve the ability of our faculty to compete for external funding.

     Because FIU is an Hispanic-Serving Institution, the attraction and retention of Hispanic, other minority, and women students is a significant issue for us. We plan to support and enhance ongoing minority programs that have already proven successful. One important objective is to create continuity between local high schools and our institution by providing qualified graduating seniors financial support and mentorship throughout their university studies. Other objectives include a full-immersion educational program for these students and a high-school teachers enhancement program -- "Innovative Educational Methods in Computer Science and Engineering." A key goal is to develop an environment for disseminating innovation in computer science, computer engineering, and related technologies. The primary target will be public schools from the inner city and other deprived areas. We also want to create a bridge to our community's public school students through an innovative pipeline program, giving them the opportunity to tap into the best educational and research resources, thereby attracting a significant number to higher education. Because FIU is one of the few minority institutions in the nation with comprehensive programs through the Ph. D. in Computer Science and Engineering, we are well positioned to meet these goals.

     CATE will collaborate with other ongoing minority student recruitment and retention efforts and create a coalition of minority institutions. Collaborative research and educational activities with major research universities in the nation are proposed and will be sought. Such collabrative efforts will be enhanced by teleconferencing and other forms of distance learning.

     We see the funds requested in this proposal as seed money. We will proactively seek permanent funding for CATE as the NSF-sponsored infrastructure matures. This support will be sought from the FIU administration, the State of Florida, and other government and private agencies, foundations, and industry. The proposal documents the potential for obtaining such support, with the understanding that the establishment of CATE is an essential pre-condition. Once established, the additional support will enable CATE to operate independently as NSF support is phased out.

The main body of the proposal will address the following issues:

1. existing educational and research environments at our institution;
2. our choice of research and educational areas, the rationale for that choice, and the relevance of the proposed infrastructure to excellence in research and education;
3. why a coherent infrastructure, as proposed, would provide much greater leverage than separate funding for each of the projects;
4. cooperation with other institutions and the importance to this proposal of interdisciplinary and interorganizational research;
5. a comprehensive plan to achieve minority participation and retention from high school through graduate school; |
6. the equipment request and justification;
7. a management plan explaining duties and responsibilities, as well as the strategy for future self-sufficiency.

     Finally, we wish to emphasize institutional commitment to the success of this project. President Modesto Maidique speaks of his "enthusiastic support for the proposal" in an accompanying letter and has decided to waive the entire overhead charge of $221,755 which now becomes matching funds. This action demonstrates our resolve to use technology and research as a foundation for serving the educational needs of the minority community

5.1 CATE COMPUTING REQUIREMENTS

     To meet both research and instructional needs, the CATE Laboratory (Fig. 5.1) provides a computing environment capable of engaging researchers as well as facilitating classroom and laboratory-based instruction for undergraduates. The demands of such academic diversity are achieved through the use of a distributed computing environment, based on client-server technology, and consisting of several powerful computing systems connected by high speed networks sharing computing resources. It is the CATE's mission to provide an innovative laboratory capable of addressing effectively the computationally-taxing problems of vision, artifical neural networks, computer-aided instruction and ISDN.

      CATE's architecture consists of: (a) parallel computational server accompanied by multi-purpose servers which disperse general processing needs, allowing the parallel server to provide computational processing; (b) graphics and file servers which are connected to a 100 Mbit network to increase the bandwidth and speed of transmission of large files such as compressed imagery and full motion video; (c) several workstations with both the capacity for computational processing and high quality graphics and sound; (d) low cost graphical systems capable of providing a responsive graphical interface for interacting with the main computing servers.

5.2 CURRENT EQUIPMENT AVAILABLE TO FACILITY

     The CATE laboratory will coexist with FIU Engineering and Computer Science laboratories which are housed in the Engineering and Computer Science (ECS) building. Present computing devices, using the ethernet, are connected to the campus computing network through fiber optics cabling to Sun Microsystems (Sun), Digital Equipment Corporation (DEC), Silicon Graphics, Inc. (SGI), Symbolics, and Inmos computing devices. Various computing systems are thus available to the CATE for remote use.

     The South East Regional Data Center (SERDAC) provides the majority of computing services on campus through the use of several Sun and DEC servers. SERDAC oversees the day-to-day operation of these and other computing facilities on campus, including data/telecommunications. Through SERDAC, FIU is a member of Internet and Bitnet and is also connected to FIRN (Florida Information Resource Network), a network of corporate and state computing systems.

5.3 PROPOSED EQUIPMENT

(1) Parallel Server-- SGI Challenge L series computer selected is configured with four MIPS RISC 4400MC microprocessors (64-bit, 100/150 MHz, 500 Dhrystones MIPS), 128MB RAM, a 1.2 GB hard drive, CD-ROM drive, 4 MM tape drive and FDDI network adapter. Expandable to 12 processors and has an internal bus throughput of 1.2GB/sec;

(2) Graphics and File Servers-- (a) The SGI Crimson VXGT computer includes a MIPS RISC 4000C processor (64-bit, 100 MHz), 64MB RAM, 3.2 GB hard drive and FDDI network adapter. The VXGT is capable of 1.1 Million Triangle-Meshed, shaded polygons per sec. Combined with the Video Lab real-time input/output frame buffer, imaging information can be analyzed and displayed in real-time; (b) The Sun SPARCserver 10 model 30 (SS10) includes a SuperSPARC processor (32-bit, 45 MHz), 64 MB RAM, 2.25 GB hard Disk, 8MM tape drive, and CD-ROM drive. The file system on the SS10 can be expanded to 41GB of SCSI based hard disk storage;

(3) Multimedia Workstations-- In its capacity as a general computing workstation, the SS10 was chosen to provide distributed processing needs, which includes multimedia and ISDN applications. The SS10, in addition to the above mentioned processor is configured with 8-bit color graphics accelerator, 32 MB RAM, and 1.2 GB MB hard disk, 16-bit audio and a dedicated ISDN processor.

(4) X-terminals-- The Network Computing Devices (NCD) MCX15, is a 15" Color (24-bit), Motorola 88100 RISC processor (20MHz/32-bit), 6MB RAM (expandable to 68MB) and 16-bit audio (input/output);

(5) Networks-- A FDDI/Ethernet network will provide the data connectivity to the main servers and other computing devices. The network will be attached to the campus ethernet which is connected by a T1 line (1.4 Mbit) to Internet. The FDDI/Ethernet hub will have a FDDI/Ethernet bridge which will isolate traffic within the center's sub-domain. Further, traffic between servers will be transmitted exclusively on FDDI. Ethernet segments emanating from the network hub will accommodate the various workstations and X-terminals. Additionally, several Basic Rate Interface connections will be acquired to provide connectivity to the ISDN ready workstations.

     The following equipment will be attached to the network directly or through one or more of the servers to support the needs of the lab: HP LaserJet IV with 8MB RAM and network card, Denning Roving Robot and Puma 260 Robotics Manipulator and Panasonic CCD Cameras.

     UNIX System V is the base operating systems for both the major vendor selected: Sun via Solaris 2.x and SGI IRIX 5.x. Both support multiprocessing, and are POSIX compliant. Both comply with Network Information System (NIS) and Network File Systems (NFS) standards. MIT's X11R5 Windowing system is available on both systems. The development languages of C/C++ for each platform will be used. Addition software listed below will be used to provide programming support: Centerline C++ software development environment, Object Works Smalltalk, IXI X. Desktop (User Interface), and Gain Technology multimedia authoring system, "Gain Momentum."

5.4 SELECTION RATIONALE

     Our main objective is to create a balanced infrastructure viable for both educational and research activities. In our selection of the equipment, the following evaluation criteria were considered important in order to address effectively the research avenues discussed in the previous sections: (a) parallel processing; (b) high performance 3-D graphics workstations; (c) real-time control; (d) use of operating system and graphics standards; (d) high speed data communications; and (d) low cost implementation. We also placed an emphasis on price to performance ratios, vendor reliability and experience with the technology, and the ability to integrate with existing campus equipment. In terms of the vendors evaluation criteria, we considered the following facts:

     The selected equipment addresses the need on campus for a center which can deliver enormous computational power. As designed, the center's processing can be distributed over 11 computing systems which house a total of 14 processors-- potentially supplying over 1.5 GIPS (1,500 Dhrystones MIPS).

     The infrastructure, as proposed, can accommodate in one setting 26 students (12 X-stations, 12 workstations (4 matching from ECE department), 2 servers) for both research and instruction with minimal bottleneck problems, while maintaining a high level of parallelism and computing performance for maximum productivity. Remotely, the center can handle well over 200 simultaneous student users from facilities which are open to all students on or off campus (estimated by on-campus and vendor technical staff, with respect to undergraduate courses in Computer Engineering). Remote use will be for instructional use, low-level computational research and other computer services for pipeline students.