Over the last ten years, the increasing complexity of the systems developed by THALES has been successfully tackled by an approach featuring:
-
Substantial upgrading of the techniques used for managing complex projects,
-
The introduction of development processes and measuring systems optimised for large systems,
-
The uniform, large-scale deployment of development support services (for better management of requirements, specification and design, testing, configurations etc.),
-
Greater investment in training.
Our ability to handle complexity now makes it possible to begin a second phase focusing on automation of the different aspects of the development and implementation of complex systems. This quest for greater automation reflects the following business considerations:
-
An increasing trend in software engineering profession to give greater importance to a capitalisable architecture,
-
Integration of system engineering and software engineering activities,
-
Redefining the client relationship, particularly as concerns maintenance and system evolvability,
-
Progressive integration of reputed and widely-supported civil technology,
-
Greater use of research and development capability and technology from academic sources,
-
A continually increasing training requirement.
Automation however raises a certain number of scientific and technical difficulties, and these constitute the focus of the research carried out within Carroll. There are, in particular, two issues that we are faced with:
-
Consideration has to be given in software development engineering to the need for increased flexibility, quality and productivity (validation at an early stage, methodologies, development tools and help facilities, flexibility of execution infrastructure, downstream testing etc.), as well as capitalisation of the development work.
-
In the field of system software infrastructure, it is necessary to address the extreme constraints associated with on-board installation, mobility, and distribution (distributed execution on a number of intercommunicating machines).
However, firstly, little use is made of tools in current software development methodology. The methodology is generally focused on a set of documentation for the system under development, that does not lend itself to automation. Opportunities for automation are also limited by the low levels of interoperability of the ready-made tools available.
Secondly, current software infrastructure is generally of limited flexibility, being too specific, and is difficult to adapt to the requirements of the types and qualities of service required for the systems planned.
This being the case, the different independent analyses lead to the conclusion that there are the following two needs:
-
It is necessary to evolve from separate, clearly identified, stable, abstract models to approaches fostering development. The range of solutions that would appear to have the best chances of success corresponds to model-driven engineering (MDE). This approach has been the subject of a substantial amount of academic work and a major campaign by the Object Management Group (OMG) designated Model-Driven Architecture (MDA). This corresponds to the first field of the Carroll Research Programme: model-driven software engineering
-
It is also necessary to upgrade the functionality and the scope of the middleware , which constitute the central part of the software infrastructure, separating the business side from the technical side. It is, indeed, the middleware technology that has continued to make it possible to raise the level of abstraction used by the architects, and thus directly address the professional requirements in the applications. This corresponds to the second field of the Carroll Research Programme: middleware technology
Last update: 07/01/2004 - Legal provisions
