SHE/POOSL Website
The purpose of this project is to provide methodological support, including (prototype) tools, for the predictable design of distributed hard real-time embedded control systems for mechatronic products. The methodology consists of three major components: views, multidisciplinary core models and correctness-preserving code generation. The views allow designers from different engineering disciplines to interactively and concurrently work on the design of the complete system, while each team member can observe and understand the impact of other one's design decisions through its own, well-known view. The different views are translated into multidisciplinary core models. The interaction between the core models will be realized by co-simulation, allowing for verification as well as for consistency checking between the different design views. Views representing software components are mapped automatically onto the target hardware platform in a correctness-preserving way. This means that the properties exhibited by the views are kept in the software realization. With this methodology, we aim to relax the tension between design cost and design time on the one hand and quality (in particular reliability and robustness) on the other hand. The methodology is tried out using several test setups. First at the university, and concurrently, but started a little later, at our industrial partners, Océ and Chess-eT.
This PhD research is performed in the context of the Boderc (Beyond the Ordinary: Design of Embedded Real-time Control) project and focusses on architecture exploration for embedded real-time control systems. The control application is described in terms of real-time communicating parallel processes. These processes are mapped onto a hardware platform consisting of microprocessors, DSPs, memories, communication buses and networks. The hardware platform and the mapping have to be chosen in such way that different functional and non-functional requirements be met. In particular for real-time control, it is important that all real-time deadlines specified for the different processes are met. The goals are twofold:
The PhD project is part of the Managing Soft Reliability in Strongly Innovative Product Creation Processes IOP project, which will provide a method and a set of corresponding models for the analysis and management of Soft Reliability Problems that can be used as integral part of the product creation process of highly innovative products like professional office products or multimedia consumer products. It addresses an important bottleneck that prevents technically feasible products from being accepted and appreciated by customers in the market.
Since one single customer complaint can have many causes inside the product methods will be developed to allow non-intrusive data logging inside the products analysed during this project. These trails will concentrate not only on the internal architecture of a system but especially on the parts of a system that interact with the outside world. This is in contrast with the traditional tracing techniques; most traditional tracing techniques (also known as debugging) concentrate on finding deviations from the formal specification during execution. This sub-project will focus especially on finding discrepancies between customer requirements and specifications and therefore a stronger focus on the product interfaces is expected. The ambition is that these activities should lead to methods that generate SRP related data early in the product creation process and within the time constraints of modern industrial product creation processes.
Currently the focus is on the general theme of self-observing products and their possible usage in various scenarios. While engaging in the design and architectural aspects of such systems, a concern is also the implementation of several real-life cases and later on the generalization of requirements that might arise in the application areas of this technique. Thinking goes into the direction of an integrated development process by using UML-based tools as well as model transformation and weaving to achieve a seamless integration of the observation aspect which is in most cases cross-cutting through large parts of the host system. On the other hand this "from-scratch" approach cannot be applied to nowadays products and product development processes so it is important to find a way to smoothly introduce observation into already existing systems. Regarding this, industrial case studies will be done.