Title: Designing Highly Evolvable System Architectures (Darwin)
Start date: 01-10-2005
End date: 30-09 -2010
Evolvability is an important property of many technical systems. Complex systems need to evolve as requirements change or new technologies become available. Typical up-scaling of requirements include changes to regulations and legislation, demands for higher performance and increased features but more generally the replacement of existing functions by improved (cheaper, smaller, lower power consumption etc.) technology.
For the systems we consider within this research project evolvability is increasingly becoming one of the major system properties that drive the business economics and competitiveness. It is essential to preserve and build upon the huge investments needed to create such systems whilst providing a continuous stream of advanced product offerings to the market.
System evolvability is a system's ability to withstand changes to its requirements, environment and implementation technologies. The need for greater systems evolvability is becoming recognized, especially in the engineering of computer based systems, where the development, commissioning and replacement of large systems is highly resource intensive. Despite this need, there are no formal means for evaluating the evolvability of a system and thus no means of proving that one system is more evolvable than another. This project recognizes this and aims to provide generic methods that will lead to the design of high evolvable systems.
The resources required to introduce additional features into highly complex systems such as MRI scanners can be significant. Even minor top level functional changes can have lengthy and difficult to predict development cycles. For these types of systems it is essential that the throughput, reliability, cost, safety and ease of use be maintained whilst providing a timely solution to the customer.
A number of trends can be seen emerging:
- Complexity increase: Solutions in response to demands for new operating features, higher performance, greater ease of use, improved accountability and safety have been increasing fast in recent years and these are driving an increase in system complexity
- Time to Market pressures: The innovation cycle of products is decreasing continuously. In most industrial sectors competition is fierce and price erosion is fast. Product manufacturers are under severe pressure to reduce product costs, reduce development costs – and yet remain technologically one step ahead of the competition
- Open Systems. The expectation of many customers is that these products can be connected to a variety of complimentary products and systems probably from different manufacturers. The product is increasingly being seen as a “function” within a much larger workflow or process. The ability to communicate and interface using open standards is essential.
When no other measures are taken, these trends lead to higher development costs and a longer time to market. Evolvability is an important systems engineering property that the whole world is struggling with. The Darwin project aims to provide the missing theoretical and technological methods for the architecting and evaluation of highly evolvable systems.
MRI scanners - an industrial project case
In our opinion, studying evolvability issues in general does not make much sense. Achieving our goals requires focus. In the Darwin project, we focus on the industrial sector of medical imaging, in particular that of MRI scanners.
MRI (Magnetic Resonance Imaging) is a technology for generating visualizations of various internal parts of the body without the use of X-rays. In the vast majority of cases MRI technology is considered extremely safe, exceptions being patients with metal implants. MRI scanners are particularly good at looking at the non-bony parts or "soft tissues" of the body. The brain, spinal cord and nerves can be seen much more clearly with MRI than with X-ray and CT scans. However, X-ray systems have the advantage that they are frequently better than MRI at viewing boney structures.
A MRI scanner is made up of a number of specialized technologies. At the heart of the machine is a very large magnet assembly which accounts for nearly 40% of the cost. Associated with this are radio frequency transmitters and receivers, signal conditioning, signal processing, display systems and data archiving function. All of these have to work seamlessly together whilst providing a safe and easy to use operating environment. Patient safety and repeatability of results are essential features of these machines.
Darwin research areas
The following research lines of attention (LoA) are followed within the Darwin project
- Systems Engineering : Determine key system drivers and evaluate the trade-offs between evolvability and system properties such as performance & cost
- Modeling and analysis: How to evaluate the impact of proposed architectural solutions on key system qualities
- Connecting systems and Software: What software architectural provisions should be designed into high-tech systems right from the start to allow easy evolvability over time
It is the objective of the Darwin project to develop methods and tools for optimizing system evolvability. I.e. the ability of a system to evolve easily. This should result in a faster time to market for product iterations whilst maximizing technology reuse.
We use four test cases from Philips Medical Systems, MRI division, to research and evaluate the proposed solutions:
- Patient support System
- Continuous moving bed imaging
- MRI as an active therapeutic tool
- Integration of a MRI scanner into the total hospital workflow
Duration, staffing, and financials
- The total duration of the project will be 4 years. The project starts October 1st, 2005 and ends September 31, 2009.
- The staffing level during this period is approximately 20 FTE on a yearly basis.
- The total project budget is approximately 8 M€. Dutch government funding under Bsik covers approximately 4 M€.
A consortium of industrial and academic partners carries out the Darwin project and the Embedded Systems Institute (ESI) manages it. For the majority of their project time, the participants are co-located at the ESI facilities in Eindhoven or during key experimental phases of the project at the Philips Medical Systems facilities in Best.
ESI hosts a secure Darwin website for exchange of documents within the project.
Participants from TRESE
- Mehmet Akşit
- Klaas van den Berg
- Pim van den Broek
- Gürcan Gülesir
- Selim Ciraçi