NeTS:Small: Collaborative Research: Predictable Delivery for Large-Scale Real-Time Applications

University Of Minnesota-Twin Cities, Minneapolis MN

Investigators

Abstract

More and more large-scale real-time distributed control and data processing systems are deployed to support the nation?s infrastructures and critical services, such as power grid control, transportation monitoring systems, and water processing systems. Due to their large scale deployment, these systems pose very different requirements, compared with traditional real-time systems. They generally require soft (predictable) real-time delivery over multiple domains on dynamic networks. These requirements cannot be met using current overprovision or reservation-based solutions. This project investigates a promising solution to effectively address this issue across multiple domains. The fundamental issues in supporting large-scale predictable real-time delivery on the Internet, by addressing the predictability in the network layer are studies here. In particular, the project explores the ready-available time synchronization technology, such as GPS, CDMA, and IEEE 1588 Precision Time Protocol, to achieve predictable delivery over multiple domains and dynamic paths. Two key mechanisms are developed to estimate the delay from a router to a destination and service a packet based on its upstream earliness and downstream laxness for more efficient resource sharing among partially-overlapping flows. The results from this project will provide a viable practical solution to support large-scale distribution control systems and enable more powerful applications in power control, traffic monitoring, and energy management. The research results will be shared with the broad community via journal and conference publications. A prototype system will be built to illustrate the effectiveness of the proposed system. This project will work with Native Hawaiian Science and Engineering Mentorship Program to broaden the participation of minority students.