Using Mission Graphs to Identify Mission Dependability Bottlenecks Abstract in Complex Systems
Abstract:
Embedded systems of today pose difficult dependability challenges. Hardware and software requirements as well as human interface components all contribute to or detract from the overall dependability of a system. Assigning a ‘dependability number ’ to a system is becoming increasingly subjective due to the confluence of these three areas. In particular it is important to go beyond composing individual component reliability predictions, and additionally consider factors such as ease of user workaround in the face of a partial system failure. In this paper we shall present an approach that attempts to detect these dependability bottlenecks within embedded systems and investigate its ability to represent users ’ ability to interact with partially failed systems. We propose a graph-based approach that is partially based on composing and extending Unified Modeling Language (UML) standards. This ‘mission graph ’ concept is meant to take advantage of the user’s perspective to help system designers understand what is really going on in complex systems. We apply this approach to an example embedded system, and examine the experimental results to determine the feasibility of the proposed approach. Finally, the mission graph approach is used to further investigate the workaround concept and how it applies to users attempting to accomplish their goals even in the face of component failures. 2 1.
