Section: Contracts and Grants with Industry

DiVA (IST )

Participants : Jean-Marc Jézéquel, Benoit Baudry, Brice Morin, Freddy Munoz, Olivier Barais, Didier Vojtisek.

The goal of DiVA is to provide a tool-supported methodology for managing dynamic variability of co-existing, co-dependent configurations in adaptive systems that span system administration and platform boundaries. Examples of such adaptive systems are communication infrastructure in rescue operations and mobile entertainment environments. This will be addressed through a combination of aspect-oriented and model-driven techniques. DiVA will explore how adaptation policies can be captured in the requirements, how aspects can model the variants used to adapt the system, how models can be kept at runtime to drive the adaptation and which validation techniques have to be developed in this context.

The Triskell team participates mainly in the definition of models that can drive the adaptation at runtime. The benefits of keeping models at runtime is to have an abstract view of the adaptation policies and mechanisms on which it is possible to reason (to check invariants, QoS properties, etc.) before actually adapting the running system. One important challenge tackled by Triskell is a mechanism to synchronize the running system with the model that has been adpated according to the changes in the environment. Triskell is also involved in the different validation tasks that occur when building such systems and when adapting these systems at runtime. An important issue for validation at design time is to select a subset of all possible configurations for testing. At design time, it is necessary to validate interactions between variants and to check that invariants on the system are satisfied.

Project duration:

2007-2010

Triskell budget share:

400 keuros

Project Coordinator:

SINTEF

Participants:

SINTEF, Uni. Lancaster, INRIA, Pure Systems, Thales IS, CAS.