End-to-End Timing Analysis of Black-Box Models in Legacy Vehicular Distributed Embedded Systems

A majority of existing techniques and tools, used in the vehicular industry, support the extraction of end-to-end timing models. Such models are used to perform timing analysis of distributed embedded systems at an abstraction level that is close to their implementation. This paper takes a first initiative to provide such a support at a higher level of abstraction. At such a level, the system can be modeled with inter-connected black-box models of nodes whose internal software architectures may not be available. However, most of the design decisions about network communication are available. This represents a typical scenario in the vehicular industry where most of the artifacts are reused from either legacy systems, other projects or previous releases of the vehicle. In this paper we present an approach for the extraction of end-to-end timing models at the highest level of abstraction used in the vehicular domain. Using these models, end-to-end path delay analysis of the systems can be performed at a higher abstraction level and at an early phase during the development. As a proof of concept we implement this technique in an industrial tool suite, Rubus-ICE, that is used for the development of these systems by several international companies. Using the extended tool, we conduct a vehicular-application case study.