Integration into the CONNECT Architecture

Abstract : The CONNECT Integrated Project aims at enabling continuous composition of networked systems, using a revolutionary approach, based based on on-the-fly synthesis of CONNECTors. The Role of Work Package 4 is to develop techniques for learning representative models of the connector-related behavior of networked peers and middleware through exploratory interaction, and for monitoring the runtime behaviour of the connected system. This document provides an overview of WP4 achievements during the final year of CONNECT, as well as a summary of WP4 achievements and remaining challenges for the entire period of CONNECT. During Y4, WP4 has further increased the power and efficiency of learning techniques, developed and implemented techniques for handling non-functional properties in learning, and finalized the integration of the learning and monitoring enablers into the CONNECT architecture. Over the 46 months of CONNECT operation, WP4 has significantly advanced the state-of-the-art in active automata learning. Prior to the CONNECT project, active learning had been developed only for finite-state component models, utilizing a finite set of interaction primitives. In CONNECT, we have lifted this technology to rich and infinite-state techniques by novel symbolic and abstraction-based techniques, thereby providing a break-through in the state-of-the-art, which will have long lasting impact also after CONNECT. During CONNECT, we have also thoroughly re-engineered our framework for learning, LearnLib, making learning functionality available as reusable components. Further, we have developed a generic monitoring infrastructure that offers great flexibility and adaptability, which is model-driven: it can thus be adapted to a rich set of domain-specific languages, expressed as metamodels, and exploit the support to automation offered by model-driven engineering techniques. Our assessment of the learning and monitoring enablers on CONNECT scenarios shows that the developed technology can cope very well and very efficiently with the challenges imposed by the CONNECT approach.
Type de document :
Rapport
[Research Report] 2012
Liste complète des métadonnées

Littérature citée [98 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-00805623
Contributeur : Emmanuelle Grousset <>
Soumis le : jeudi 28 mars 2013 - 14:14:26
Dernière modification le : mercredi 4 octobre 2017 - 11:24:02
Document(s) archivé(s) le : samedi 29 juin 2013 - 04:03:55

Fichier

CONNECT_Deliverable_D4_4.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-00805623, version 1

Collections

Citation

Antonia Bertolino, Antonello Calabro, Sofia Cassel, Yu-Fang Chen, Falk Howar, et al.. Integration into the CONNECT Architecture. [Research Report] 2012. 〈hal-00805623〉

Partager

Métriques

Consultations de la notice

127

Téléchargements de fichiers

133