Integrating dependability analysis and optimization into the distributed embedded system design process

The goal of this research is to add dependability analysis into the design of distributed embedded systems. This helps system designers evaluate system dependability, while other system evaluation concerns, such as system performance, and design cost, are analyzed during every step in the system design process. Currently, there exists very few embedded system design frameworks that consider both dependability and performance properties of all the software and hardware components together during the design process, and nave of them support design for all design levels in the system design process, including evaluations of system redundancy, and dependency in failure behavior of the components that have both software and hardware. To solve these problems, we have designed a framework for integrating dependability analysis into the design of distributed, heterogeneous, fault-tolerant embedded systems. Our system dependability analysis provides choices of system design based on the dependability results. In addition, we perform system dependability evaluation and optimization early in the system design process, without needing complete design information. In other words, with incomplete design information, we are able to predict the behavior of system dependability. This benefit significantly reduces time and cost in system design. We present several models to demonstrate our techniques for optimizing system reliability and cost for fault-tolerant distributed embedded systems. The models help us find the optimal embedded (software and hardware) system structures while considering basic information on reliability, cost and weight of the components. Each model is suitable for a distinct set of conditions or situations. The models maximize reliability while meeting cost and/or weight constraints, or minimize cost while meeting reliability constraints.