PreView: Developing Predictable Vehicle Software on Multi-core

In this project we aim to develop model- and component-based software development techniques for on-board vehicular control systems, having multiple criticality levels, to be deployed on multi-core platforms. The techniques will support various development steps, i.e., from modeling of the software architecture to its timing analysis, synthesis and predictable execution on multi-core platforms. Multiple criticality levels in the vehicle software will be supported by means of partitions in the core(s) of single-core as well as multi-core platforms. Among the above-mentioned steps, the project will put high focus on supporting predictable execution of these systems on such platforms. Hence, an end-to-end timing analysis framework will be developed to verify predictable timing behavior of these systems. Moreover, in order to provide a predictable run-time support for these systems on multi-core platforms, we aim to develop a virtualization technique that supports the reuse of a certified single-core Real-Time Operating System (RTOS) by means of a multi-core hypervisor.

Our goal is to use our industrial partners to make industrialization of research results that are obtained from this project. For this purpose, we will develop a proof of concept demonstrator by implementing these techniques in existing commercial models and tools that are actually used by the industry. In particular, the modeling and synthesis techniques will be implemented in the existing industrial model, the Rubus Component Model and its tool suite Rubus-ICE. The end-to-end timing analysis will be implemented as a plug-in for Rubus-ICE. Currently, the Rubus RTOS supports only single-core platforms. It has already been certified in ISO 26262:2011 safety standard. The newly developed multi-core hypervisor will reuse a separate instance of the Rubus RTOS per core. The efficacy of the extended Rubus tool suite will be demonstrated and evaluated on the industrial use cases that will be provided by our industrial partners. In summary we believe that the results of this project will be instrumental when addressing key challenges in the development of next-generation complex vehicular embedded software systems.