Clinic Projects

Please click on a link below to view the Harvey Mudd College Computer Science Clinic projects for the corresponding time period.

Clinic Projects for 1998-1999

Visualization of Data with Uncertainty

Client
HRL Laboratories, LLC

Faculty Advisor
Professor Ran Libeskind-Hadas

Student Team
Jascha Swisher (Team Leader), Denny Chan, and Alan Hatakeyama
While visualized data is often not of the highest quality, traditional methods of scientific visualization typically ignore these inherent uncertainties, leading to a potentially misleading representation. The Clinic project called for developing an application demonstrating various means of accurately representing uncertainty using the newly-released Java 3D API.

Extending Active Calendar

Client
IBM Almaden Research Center

Faculty Advisor
Professor Josh Hodas

Student Team
James Courtney (Team Leader), Bing Chen, Beth Gordon, and J. Nathaniel Sloan
The IBM Active Calendar group has developed a prototype system capable of examining the contents of a user's calendar (when stored in a client-server system such as Lotus Notes) and inserting information relevant to individual appointments found on the intranet and internet into the entries. The goals of this Clinic project were to add support for the Microsoft Exchange/Outlook calendaring system, and to propose and implement a re-architecturing of the system designed to significantly increase efficiency and scalability.

Improving Network Backups by Automated Analysis

Client
Legato Systems, Inc.

Faculty Advisor
Professor Margaret Fleck

Student Team
Tom Vaughan (Team Leader), Tyler Daenzer, Chris Lewis, and Wayne Yang
Companies with large networks of computers face major challenges in quickly and reliably backing up their data. Legato's NetWorker software provides a solution to this problem by coordinating backups across a network. While basic installation and backup configuration is simple, optimization of the NetWorker software is more challenging. To increase the functionality and ease-of-use to Legato's customer, the Legato Clinic team analyzed the NetWorker software product and implemented a solution that will allow NetWorker users to quickly and easily optimize their NetWorker configuration.

Algorithms for Router Table Generation and Access

Client
Nuera Communications, Inc.

Faculty Advisor
Professor Ran Libeskind-Hadas

Student Team
Janet Davis (Team Leader), Ben Barden, Pascha Karimov, and Matthew Wright
Nuera Communications produces switches that enable corporations to route telephone calls between corporate sites over leased lines and the internet, thus reducing telephony costs. The Clinic project for 1998-1999 had two major goals. The first was to automatically generate efficient routing tables for use in Nuera's telephony devices, thus eliminating one of the most difficult and error-prone steps in configuration. The second was to simplify the process of testing Nuera networks by simulating them in software.

Use and Extension of a Software Requirements Specification Tool

Client
Rockwell Science Center

Faculty Advisor
Professor Wing Tam and Dr. Jon Strauss

Student Team
Andre Abramenko (Team Leader), David Browne, James Holloway, and Jim Tran
Prefer is a system requirements specification package developed at RSC. During the first semester, in order to become familiar with the system, the Clinic team used it to specify requirements for a car access control system of the future. During second semester the goal was to investigate the practicality of using CORBA as a foundation for re-architecting Prefer as a distributed multi-user client-server application.

Network Intrusion Detection

Client
The Aerospace Corporation

Faculty Advisor
Professor Mike Erlinger

Student Team
Jeff Lawson (Team Leader), Nathan Field, Glen Gebhart, and Dave Meyer
The field of Intrusion Detection has grown nearly as rapidly as our reliance upon computers and the need to keep their data secure. Unfortunately current automated intrusion detection systems are inherently error-prone and inaccurate. The Aerospace Clinic team tryed to improved the accuracy of intrusion detection methods by using data correlation techniques to reduce false positives and to associate separate, seemingly harmless events with actual computer attacks.

A System for Managing Time-Based Thresholds

Client
WebTV Networks, Inc.

Faculty Advisor
Professor Geoff Kuenning

Student Team
Peter Boothe (Team Leader), Seth Heidkamp, and David Minor
The WebTV internet service is supported by a large network of computers and network devices. At the time of this project the networking group collected operating data from a large number of these devices, but makes little use of it because there is too much data and it was hard to tell what is right from what is wrong. The goal of this Clinic project was to build a tool to generate time-based performance thresholds to detect from the data whether devices are behaving correctly.