Composable Embedded Systems for Healthcare
Project Number: P11
Project Manager: Prof. dr. ir. B.R. Haverkort / Dr. ir. Aiko Pras
Faculty of Electrical Engineering, Mathematics and Computer Science
The major scientific aim is provide a solid mathematical basis for the design and analysis of highly-adaptable component-based software systems with critical performance and safety constraints. Given the objective to support the trade-off between several system qualities, multi-dimensional models will be developed that can be projected to various system views and analysis techniques.
The key element of this approach is the definition of a formal model of a software component. The main motivation for this choice is that it provides the common basis for all analysis techniques and hence supports the integration of these techniques. A component is also the link between high-level system views and more fine-grained design. Component interfaces will be structured according to system qualities and by using restricted views specific aspects of the design can be analyzed.
An essential part of the approach is the validation of scientific results on representative industrial-scale examples. In this respect we follow the industry-as-lab philosophy that has been used successfully in earlier Bsik projects led by ESI. It avoids the drawbacks of the classical research-then-transfer approach where transfer to industry is attempted when the research has been completed. It may turn out that industrial applicability is complicated (e.g., because unrealistic assumptions have been made). In the industry-as-lab approach immediate feedback is obtained allowing quick adaptations of research results to be made. It often leads to new research challenges.
In this proposed project, the contribution of the industrial partners is critical because realistic experiments on industrial-scale archives are not practical in a purely academic setting. Philips Healthcare will also provide knowledge about the specific requirements used in this demanding application domain, such as safety and performance requirements.
Research on the questions mentioned in the previous section will result in significant new scientific results such as:
- Multi-dimensional models that allow reasoning about multiple system qualities
- Architectural guidelines for evolvable embedded software that also take safety and performance into account
- Formal component models that allow multiple view attributes to be analyzed
- Suitable constraints on architectures and components that facilitate compostional reasoning about system properties
- Improved performance analysis techniques that allow decomposition and trade-offs between analysis speed and accuracy
- Analysis results on the balance between evolvability and performance constraints
- Techniques to reason about safety and performance for large numbers of software configurations
- Validation of the developed academic results using industrial examples
Besides the ESI project Darwin, related projects can be found in the NWO research program Jacquard. Of particular relevance are the projects TraCE (Transparent Configuration Environment), which emphasizes composition of components, and MODSE (Model-driven software evolution). In the European context, results of the GENESYS project on the definition of a GENeric Embedded SYStem Platform can be used. Also relevant is the formal framework for component-based design of complex embedded systems that will be developed in IST FP7 project COMBEST. Interesting are IST projects on general modeling notations for embedded systems such as ADAMS, on a UML profile for modeling and analysis of real-time and embedded systems, and SATURN which studies the use of SysML for modeling, architecture exploration, simulation and synthesis for complex embedded systems.
Project duration: 1-9-2011 – 1-9-2016
Project budget: 6.1 M-€ / 2.7 M-€ funding
Number of person/years: 541 person/months
Project Coordinator: Philips HC
Participants: Philips HC, Axini, TU Delft, UT, ESI
Project budget CTIT: 504 k-€ / 226.8 k-€ funding
Number of person/years CTIT: 60 person/months, 1.5 fte / yr
Involved groups: Design and Analysis of Communication Systems (DACS)