Research Experiences for Undergraduates in Dynamic Distributed Real-Time Systems
REU Projects

Summer 2000:

• REU 2000 workshop (Real Video)

• REU 2000 project report

• REU 2000 project presentation slides

• Distributed Resource Management for Agent Based Real-Time Mission-Critical Systems (Kshiti Desai, Chris Forrest, and Geoff Dale, Summer 2000)
  
  With the evolution toward agent based computing in real-time mission-critical systems which function in dynamic environments, it is important to unify the agent based computing paradigm with the theory of dynamic real-time systems. In this project, we are synergistically combining DeSiDeRaTa's model of dynamic real-time systems with agent based computing models. Agent based computing offers the ability to decentralize computing solutions via incorporation of autonomy and intelligence into cooperating, distributed applications. While this approach alone affords many benefits, its effectiveness can be enhanced by combining it with system-level resource allocation optimization approaches. The result will be a hierarchical system management capability that incorporates agent-level decision making (based on individual agent goals) as well as system-level resource and quality-of-service management (based on mission goals). Thus, the project will produce a resource management model of dynamic real-time agent based mission-critical systems, and will employ the model to develop a set of tools for adaptive resource management of agent based systems. Our results will be validated initially with a distributed dynamic real-time benchmark suite called DynBench. Transition to robotic systems will also be pursued actively.

• Adaptive Management of Computing and Network Resources for Spacecraft Systems (Ryan Detter, Summer 2000)
  
  It is likely that NASA's future spacecraft systems will consist of distributed processes which will handle dynamically varying workloads in response to perceived scientific events, spacecraft environments, spacecraft anomalies and user commands. Since these various states can not be determined prior to deployment, a method to dynamically adapt the system to handle the dynamic conditions was needed. To address this, we are evolving the DeSiDeRaTa adaptive resource management approach to enable reconfigurable ground and space information systems.This approach embodies a set of middleware for adapting resource allocations, and a framework for reasoning about real-time performance of distributed application systems.We are integrating the DeSiDeRaTa adaptive resource management system with a ground-based testbed called ITOS, Integrated Test and Operations System, that will enable NASA to perform early evaluation of adaptive resource management techniques without the expense of deploying the systems in space. NASA is evolving the Principal Investigator's DARPA-funded DeSiDeRaTa adaptive resource management approach to enable reconfigurable real-time ground and space information systems.The benefits of the effort are numerous, including the ability to use sensors in new ways not anticipated at design time; the production of information technology that ties intra-constellation networks together; the accommodation of greater numbers of missions with fewer resources.

Summer 1999:

• Distributed Research Management for Multi-Robot Automated Mapping of an Unknown Environment (Michael Youngblood, Summer 1999)
  
  The Parallel Intelligent Knowledge-building Exploration program (PIKE) was originally designed and written by Michael Youngblood as an undergraduate honors project at UTA. The purpose of the PIKE system is to give an autonomous robot the ability to map an unknown environment. Placing the robot in an arbitrary position in a room (unaware of its position), the robot will create a map by exploring the environment and collecting information using sonar readings. The system is designed to run with physical robots or in a simulated environment. The purpose of this REU project is to integrate PIKE into the DeSiDeRaTa system. The goal is to modify this PIKE application to be scalable and survivable as supported by the DeSiDeRaTa environment. In this project we focused on the simulated version of the PIKE system. This system consists of three executables: Egrid (the primary application), Saphira (a commercial application launched by Egrid to handle communication with the robot), and Pioneer (a commerical robot simulator). An additional program, Sentinel, was written by Michael to provide text-to-speech audio output of the operations and status of the robot, but we did not include this program in our work. The Egrid manager was written and integrated into the DeSiDeRata system. This manager was successfully run using 2, 3, and 4 robots all working simultaneously with synchronized maps.