The industrial project partners Airbus and Bosch have vast experience in the development of time-critical systems and in the application of timing-analysis technology. They will provide the research partners with challenges in industrial system design, in addition to confronting them with the most likely developments of system design in the future.
Airbus will pose challenges to the PREDATOR partners resulting from its vast experience with the design of time-critical embedded systems. These challenges will in particular be in areas where time-unpredictability was a major concern in the past. The challenge to the academic partners will be to identify the cause of the unpredictability of execution times and to come up with improved methods to eliminate these causes. Airbus will not only look back to past designs, but will also consider likely future system architectures, taking into account ever increasing processing power needs.
Some Airbus avionics computers require hard evidence of the correctness of safe WCET estimates. Of course, efficiency is desirable, i.e. the closer WCET estimates are to actual execution times, the better. Despite Moore’s Law, it is not clear that demonstrable performance per processor will increase forever. Hence, due to to their good energy vs. performance ratio and their scalability, discrete multi-processor, multi-core and multi-processor system-on-chip (MPSoC) architectures are the most promising candidates.
Airbus will describe use cases for critical avionics computer systems in terms of hardware and software architecture. Airbus will present elements of its past designs relevant to PREDATOR objectives, and experienced difficulties.
Notwithstanding other fields of applicability for PREDATOR results, Airbus will offer guidance to ensure PREDATOR work is aligned with avionics requirements and emphasise on regulatory and industrial aspects of WCET requirements.
In the automotive domain, time-critical and non-time-critical systems are often combined in the same network or even on the same ECU. Hardware costs are an essential factor for the large volume market. On the one hand, reducing of resource consumption is an important enabler for business success. On the other hand timing problems in late development phases lead to high costs and reduce flexibility for software modifications or extensions.
Bosch will provide typical application scenarios in the automotive industry with different timing constraints. Furthermore, experience with handling unpredictability is gathered from Bosch’s different automotive business units. The saving potential and the reduction of possible risks should be carefully analysed together with the academic partners. From these application scenarios use-cases that would profit from predictable system design and system requirements should be elaborated. This includes new business opportunities by using predictable multi-core, multiprocessor, MPSoC architectures.