Achieving and Validating Higher Dependability Through Host and Network Processor Collaboration

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

The purpose of the proposed project is developing novel fault-tolerance techniques which exploit the synergy between the host and network processors in high-speed networked systems, and providing tools for their validation and the assessment of their effectiveness. Our project will include the following tasks: * Construct a layered architecture with a symbiotic relationship between the network and host processors so that fault tolerance is provided. * Develop expanded fault injection capabilities in order to study and fully understand the fault propagation and detection process in a network architecture. * Develop a tool for validating our fault detection and recovery policies and evaluate their effectiveness in a networked system using both traditional and specially constructed dependability measures. * Integrate our fault tolerance layer and the evaluation and validation tool into the NASA testbed. This project will deliver techniques to better utilize the resources of modern network architectures for fault detection and recovery. Such techniques are crucial in space missions, which are characterized by the need for a proper management of computational and networking resources for missions lasting years in severely resource-constrained applications.