Dependency modeling and phase analysis for embedded computer based systems

摘要

This dissertation presents a general methodology for modeling and analysis of multiple phased systems and a taxonomy for the solution of the resulting models, with a focus on embedded computer based systems. An example of such a system is an avionics system that includes phases such as takeoff, ascent, cruise and descent. The failure rates of components, system configuration and success (failure) criteria may vary from phase to phase. However, the failure behavior of the system during each phase is dependent upon the behavior of the system in the previous phases.;We present a unified framework that organizes the attributes of multiphased systems into four orthogonal attributes. Clear identification of each of these attributes allows us to understand, define and specify the multi-phased system. Once the system has been specified, we can explore it further and determine the solution technique that is applicable to it.;The analysis of the framework and the current approaches revealed a key omission. Conventional approaches looked at particular cases where every phase is static, or every phase is dynamic, but there did not exist a general methodology for mapping between phases, especially if the phases were of different types. We present a general interface definition and methodology to map between phases. It can be shown that the current approaches are special cases of our general methodology. We present a taxonomy for solving the different combinations of static and dynamic phases in a multi-phase system.