SMARTCore: Model-driven support for deployment optimization of embedded systems on multicore

Status:

finished

Start date:

2014-09-01

End date:

2016-08-31

Thanks to continuous advances in both software and hardware technologies the power of modern embedded systems is ever increasing along with their complexity. Among the others, Model-Driven Engineering (MDE) has grown consideration for mitigating this complexity through its ability to shift the focus of the development from hand-written code to models from which the implementation is automatically generated.
Generally MDE is meant to reduce final product verification and validation effort and cost, as well as shorten time-to-market, by providing correctness-by-construction, which opposes to the more costly correctness-by-correction typical of code-centric approaches. The way towards correctness-by-construction is often undermined by the inability of current MDE approaches to preserve certain extra-functional properties such as CPU and memory usage, execution time and power consumption. These properties are in fact, especially in the embedded domain, extremely hard or even impossible to be accurately predicted at modelling level without code execution. This is due to, e.g., resource limitation and sharing, which are common characteristics of embedded systems. This issue is amplified when approaching multicore solutions where the intricacy of diverse deployment choices leads to challenges in accurately predicting control and data flows. With SMARTCore we address several of open challenges focusing on multicore embedded systems. The aim is to provide (i) a framework which provides tangible benefits and therefore is legitimately eligible for industrial adoption, and (ii) a solid cornerstone for future academic research in model-driven engineering.
Towards the achievement of the project goals we identified the following research challenges to be tackled:
- Automatic generation of optimized code from system models for deployment on multicore;
- Monitoring code execution for selected extra-functional properties;
- Back-propagation from runtime to models for deployment optimization at modelling level;
- Runtime adaptation for resources optimization.
The expected result of SMARTCore is a model-driven framework for multicore embedded systems which tackles the aforementioned challenges.
This project brings together world leading competence in software engineering, model-driven engineering and resource optimization for embedded systems (Mälardalen University), and market leading expertise in development of embedded systems in different business areas (ABB Corporate Research, Ericsson AB, Alter Sverige AB). These project members will work tightly together to provide useful industrial solutions and novel high-quality academic impact.

PartnerType
ABB Corporate Research Industrial
Alten Sverige AB Industrial
Ericsson AB Industrial