Reliability Modeling of Dependable Distributed Systems.

The overall goal of this project is to develop theory for improved reliability modeling of systems with mixed fault types. This also provides the basis for formal methods to be used in the specification design, construction and verification of ultra-reliable multiprocessor systems. We assume that a fault is an anomalous physical condition, the identified or hypothesized cause of an error, which may eventually lead to a failure, a loss of service. In the initial project phase, our goal is to develop a hybrid fault and static reliability model that addresses mixed fault types. An in depth study of faults, including their sources, their manifestations, and the techniques needed to reduce malicious fault effects, will then provide accurate inputs to the hybrid model. We partitioned our work into three tasks analysis of static reliability models, investigation of faults, and evaluation of the impact of fault containment on fault effects and on system reliability.