Allegio - Composable Embedded Systems for Healthcare
Funded by ICTRegie
Industrial practice shows that it is extremely difficult to keep improving devices by adapting and extending the embedded software. Small changes often lead to unexpected consequences, due to unknown interactions and dependencies between parts of the system. The complexity of millions of lines of legacy code makes it almost impossible to predict the impact of changes. The replacement or integration of new software components often leads to errors that become visible only late in the software development process. This in turn causes a long and unpredictable time to market. Projects can significant exceed their budget and cancelled projects are no exception in the domain of embedded systems.
The objective of this project is to define an integrated set of methods and techniques that supports a systematic “Right by Design” component-based software design approach for complex embedded systems. This ensures that incremental changes become predictable and meet the required system qualities such as performance and safety. The Allegio project addresses the following key research questions:
- How to decompose systems into components such that the global system requirements can be derived from the properties of the components? What are appropriate coherent systems views to reasons about various system qualities such as evolvability, performance and safety?
- How to define interfaces of components that enable verifiable composition, re-use, configuration management, and replacement? How to include evolvability and safety aspects and how to characterize performance in a modular way?
- How to reason about correctness aspects of large complex embedded systems. How to analyze the trade-off between system qualities early in the design process?
- How do model-based verification techniques relate to model-based testing? What is the relation between the models that are suitable for these techniques? Does the verification reduce the testing effort? Can the testing results improve the scalability of the model checking techniques?
The contribution of SERG/TU-Delft is on investigating means to enable "Evolvable Software Architectures" for embedded systems in the healthcare domain. In particular, our research comprises (but is not limited to):
- Evolution of large, complex embedded systems
- Interface complexity and quality anlysis
- Change impact analysis and prediction
- Executable architectural models
- March 2012 --- Nicolas Dintzner started his PhD research in the Allegio project. Welcome to Allegio and the Netherlands!