Skip to Main content Skip to Navigation

Modèle conceptuel uniforme pour l’adaptation des agents logiciels en environnement ambiant

Abstract : This work deals with adaptation issues of computer systems that operate in an ambient environment. Such system is characterized by a set of heterogeneous components, which are distributed and connected to a network. The ambient environment is highly dynamic. Unpredictable events may result in a potential disturbance at any time. Therefore, on the one hand, the cohesion of their collective activities is not always guaranteed. On the other hand, the normal functioning is not always maintained when disturbances occur. To cope with these issues, we propose a conceptual model of architecture that provides adaptation. The point is to optimize their collective activity. Our main contribution is a uniform and generic pattern. We call it GMAS. It relies on a behaviour-based approach. The special feature of GMAS is the checking of the applicability of a behaviour before its execution. Checking criteria are mainly the current state of both the external surroundings and the action interfaces. Future behaviour is adjusted according to checking outcome. This mechanism leads to adaptation. The GMAS architecture model enables independent management of behaviours and exceptional events. As a result, the objectives of the agent or those of the collective do not have to take into account these exceptional events in their source code. This model uniformly manages adaptation issues of both the individual level and the collective level. We argue that collective adaptation comprises the consideration of the individuality. We aim to make agent flexible once it is at a collective level. To this end, we instantiate GMAS to design an agent internal architecture for the individual level. For the collective level, we instantiate GMAS as a coordination model for collective behaviours. This instantiated model is based on the distribution of the operating modalities.With these three conceptual models: generic, for individual level and for the collective level, we prevent harmful situations during the execution. We implement each model as a Java library to validate its effectiveness.
Document type :
Complete list of metadatas

Cited literature [70 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Wednesday, February 5, 2020 - 3:57:22 PM
Last modification on : Monday, July 13, 2020 - 9:44:02 AM
Long-term archiving on: : Wednesday, May 6, 2020 - 5:13:28 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02468232, version 1


Irène Velontrasina. Modèle conceptuel uniforme pour l’adaptation des agents logiciels en environnement ambiant. Informatique. Université de la Réunion, 2019. Français. ⟨NNT : 2019LARE0015⟩. ⟨tel-02468232⟩



Record views


Files downloads