Project Team Adam

Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry
Partnerships and Cooperations
PDF e-pub XML

Section: New Results

SCeSAME: Formal Definition of Software Architecture Adaptation

Participants : Rubby Casallas, Laurence Duchien, Nohra Villegas, Gabriel Tamura.

In order to define properties on adaptation process, we need to formally model the architecture reconfiguration of a component-based (CB) system as an action performed by itself. These actions are performed in response to the disruption of Quality of Service (QoS) contracts, in the spirit of the Effeil's rescue clause in object-oriented programming. By doing this, we aim to develop on the vision of the CBSE as a sound base to produce software systems enabled to automatically and safely reconfigure themselves by reconfiguring their abstract (reflection) architectures at runtime. For such structural reconfigurations, a system architect may reuse design patterns from other disciplines with the purpose of restoring QoS contracts, thus preserving them.

Our approach, named SCeSAME for “A Safe Contract-based Self-Adaptive Framework to Preserve QoS Properties on Mobile Devices” is built on the theory of extended graph (e-graph) rewriting proposed in e-graph  [60] , as a formalism to represent QoS contracts on component. We have given a formal definition of component-based structure systems, QoS contracts, and architecture reconfiguration rules. Based on these definitions, we built a framework that enables a component-based system to preserve its QoS contracts through architecture self-reconfiguration as a responding action to QoS contract violations. Our approach extends a theory of graph rewriting and defines a process calculus as formalisms to model the structure and reconfiguration process of architecture reconfiguration. The reconfiguration process, once parameterized with reconfiguration rules, can be verified as safe, i.e., component structural-compliant, terminating and confluent. This result is a part of Gabriel Tamura's PhD and the results have been published in [36] , [44] .