| A dynamic distribution and load balancing experiment with synchronous programming-based mobile objects | |
| référence : | DOA2003, International Symposium on Distributed Objects and Applications, 3-7 novembre 2003,, Catania (Italie), LNCS 2888, Springer, p. 1188-1207. |
| co-auteur : | Laurent Hazard |
| Transparently changing the distribution of objects in a distributed system (application, service) at runtime is a way to improve service continuity and performance. For instance, it enables to transparently shutdown execution nodes for necessary maintenance operations. Moreover, optimizing service execution resources consumption, while improving service response time, is a key to both client and service provider satisfaction. However, previous work on load balancing has globally shown practical difficulties in getting effective such an approach. In the context of advanced telecommunication service execution environments, this paper proposes to revisit preemptive dynamic load balancing, by showing an experiment based on the combination between synchronous programming-based reactive objects, a flexible object request broker and a mobile agent framework. We present a full Java-based architecture, including a service example, and we show how peculiarities of synchronous programming give a chance to support low-cost preemptive mobility (also known as strong mobility), and to easily give a relevant computer-independent load indicator. | |
| CLIF is a Load Injection Framework | |
| référence : | Middleware Benchmarking: Approaches, Results, Experiences - OOPSLA 2003 Workshop, 26 octobre 2003, Anaheim (Californie, USA). |
| co-auteur : | Emmanuel Cecchet (INRIA Rhône-Alpes) |
| This paper presents the architecture and preliminary development of Clif, a new project initiated by the ObjectWeb consortium in the field of benchmarking. Clif intends to provide a generic platform suitable for load injection and performance measurement on any system/protocol, using arbitrary load scenarios/profiles. As of current status, Clif consists of a distributed Java framework based on ObjectWeb's Fractal component model. | |
| Routing Information In Distributed Knowledge Environments With Reactive Mobile Agents | |
| référence : | PDPTA2003, International Conference on Parallel and Distributed Processing Techniques and Applications, actes de 2003 International MultiConference in Computer Science and Engineering, Las Vegas, juin 2003, H.R. Arabnia éditeur, CSREA Press, ISBN 1-892512-40-8. |
| auteur principal : | Salah El Falou (Université de Caen) |
| co-auteur : | François Bourdon (Université de Caen) |
| Static routing, is an optimization problem, which can be solved when the network topology has been defined. Large scale networks, like Internet, have to face unpredictable behaviours: trafic conditions are constantly changing, the structure of the network itself may fluctuate (switching stations or connections can fail), ... In that context, dynamic routing systems seem to be a good approach. Some of them can be seen as quasistatic, others are highly dynamic, most of them are centralized, which increases the bandwidth according to information exchange for the network control. We propose a decentralized routing protocol with mobile reactive agents for network control, based on ant algorithms, where, at the beginning, clients and servers know neither the form of the network, nor the place of others agents on the network. We introduce a capacity model for information evaporation through the network, and a Time To Live (TTL) parameter which indicates how long a packet should be allowed to survive before it is discarded. We discuss the roles of those parameters in the self-organizing process, making the search of way on the network as best as possible (with regard to QoS) during time. | |
| Programming and Executing Telecommunication Service Logic with Moorea Reactive Mobile Agents | |
| référence : | MATA2002, 4th International Workshop on Mobile Agents for Telecommunication Applications, Barcelone, octobre 2002, Lecture Notes in Computer Science 2521, Springer, p. 48-57. |
| co-auteurs : | Anne-Marie Tagant, Laurent Hazard |
| In the context of advanced telecommunication service execution environments, this paper presents a reactive mobile agent platform, based on a synchronous programming model, a flexible object request broker, and OMG's MAF specifications on mobile agent systems interoperability. The paper details the specific design of the platform, and shows how it addresses key issues such as scalability, reliability, transparent mobility and interoperability. Finally, the paper shows the use of Moorea agents for executing telecommunication service logic in the ITEA Athos European project's Enhanced Call Server architecture, featuring transparent and dynamic distributed system reconfiguration. | |
| Les agents mobiles réactifs Mooréa - une approche réactive pour la transparence à la mobilité et le passage à l'échelle | |
| référence : | revue RSTI série TSI Volume 21, no 6/2002 "Agents et codes mobiles", Hermès Lavoisier, p. 851-876. |
| co-auteurs : | Anne-Marie Tagant, Laurent Hazard, Huan Tran Viet |
| Les technologies à agents mobiles offrent un modèle de programmation unifié de la mobilité de code, de contrôle d'exécution et de données. La mise en œuvre de ce modèle implique l'implémentation d'un modèle d'agent particulier, associant activité autonome et communications (entre agents et avec l'environnement d'exécution). Or, le choix de ce modèle et des outils d'implémentation (e.g. langage de programmation, intergiciel) détermine deux propriétés non fonctionnelles importantes : (1) la capacité d'assurer une totale transparence à la mobilité pour un coût raisonnable, et (2) l'aptitude au passage à l'échelle en terme de nombre d'agents simultanément exécutables dans un environnement donné. Au travers de la présentation de la plate-forme à agents mobiles Mooréa, nous montrons comment l'utilisation d'un modèle d'agent réactif inspiré de la programmation synchrone apporte une réponse particulièrement pertinente à ces deux problématiques. | |
| Moorea, a Service Execution Environment for Telecommunication Application | |
| référence : | 7th International Conference on Intelligence in next generation Networks. Bordeaux, octobre 2001. p. 49-54 |
| co-auteurs : | Anne-Marie Tagant, Laurent Hazard, Huan Tran Viet |
| In the context of advanced telecommunication service execution environments, this paper presents a reactive mobile agent platform, based on a synchronous programming model, a flexible object request broker, and OMG's MASIF specifications on mobile agent systems interoperability. The paper details the specific design of the platform, and shows how it addresses key issues such as scalability, reliability, transparent mobility, and interoperability. Finally, the paper explains how telecommunication service execution environments may benefit from these advanced features. | |
| MobiliTools: An OMG Standards-based Toolbox for Agent Mobility and Interoperability | |
| référence : | SmartNet2000, 6th IFIP Conference on Intelligence in Networks, Vienne, septembre 2000. Kluwer Academic Publishers, p. 353-366. |
| One of the keys to success for applications of mobile and/or intelligent agents in large-scale open systems such as Internet is the ability of heterogeneous agents to cooperate and negotiate, and meet if they are mobile. This heterogeneity support requires the adoption of standards at the underlying distributed system level to support interoperability in agent management, mobile agent transport, and agent communication transport. This paper shows how both OMG standards and a modular architecture based on three kinds of component - agent mobility kernel, agent communication tools, and agent activity kernel - makes it possible to build a variety of heterogeneous mobile agent platforms with ad hoc features while preserving interoperability. | |
| MobiliTools: A Toolbox for Agent Mobility and Interoperability Based on OMG Standards | |
| référence : | poster ASA/MA2000 (2nd International Symposium on Agent Systems and Applications, 4th International Symposium on Mobile Agents), Zürich, septembre 2000. |
| One of the keys to success for applications of mobile and/or intelligent agents in large-scale systems such as Internet is the ability of heterogeneous agents to cooperate and negotiate, and meet if they are mobile. This requires the adoption of standards for the underlying distributed system to support interoperability in agent management, mobile agent transport and localization, and agent communication transport. This poster introduces MobiliTools, an OMG standards-based solution to these issues, and show how mobile agent platforms can be built from the combination of three independent components, respectively addressing mobility, communication, and activity concerns. | |
| Towards Full Agent Interoperability | |
| référence : | 2nd International ACTS Workshop on Advanced Services in Fixed and Mobile Telecommunications Networks. Centre for Wireless Communications, Singapour, septembre 1999. |
| co-auteur : | Huan Tran Viet |
| This paper introduces a CORBA service dedicated to the transport of agent communication. It explains how this service completes the emerging standards in agent technology to enhance interoperability, while supporting peculiarities of mobile agents. It also shows how this service may be customized to support a variety of message-based agent communication models. | |
| From Interoperability to Cooperation: Building Intelligent Agents on Middleware | |
| référence : | 2nd International Workshop on Intelligent Agents for Telecommunication Applications. Paris, juillet 1998 |
| As agent technologies are increasingly being involved in telecommunication-related applications, the need for open standards is becoming critical. During the past years, different scientific communities gave birth to different standardization actions, such as the Foundation for Physical Intelligent Agents (FIPA) and the Object Management Group's MASIF (Mobile Agent System Interoperability Facilities). Although they finally share some major targets, the OMG and FIPA current results show their distinct origins, respectively with a Distributed Artificial Intelligence and Multi-Agent Systems awareness on the one hand, and a telecommunication and information technologies background on the other hand. In a context where these two actions think about joining their achievements to upgrade each other, this article reports several experiments, carried out during the five past years in the agent platforms field, mixing both the intelligence and the middleware aspects. | |
| Modélisation de la coopération et de la synchronisation dans les systèmes d'information - une expérience de workflow basée sur les nouvelles technologies | |
| référence : | Calculateurs Parallèles Volume 9 / No 2 - juin 1997. Hermes, ISSN 1260-3198, p. 183-207. |
| co-auteur : | François Bourdon |
| La gestion des systèmes d'information des entreprises se doit de tenir compte de certaines technologies nouvelles, qui illustrent à la fois une ouverture de plus en plus grande de ces systèmes vers le monde extérieur, et une complexification de leur fonctionnement. Dans ce travail, nous proposons d'une part une modélisation de la synchronisation et de la coopération de processus administratifs ("workflow"), et, d'autre part, un cadre architectural et technologique capable de supporter la réalisation d'une telle modélisation. Enfin, nous illustrons cette approche en décrivant une expérience réalisée dans le cadre du projet PREVISIA. | |
| Une approche multi-agents des systèmes de bureautique communicante | |
| référence : | Thèse de doctorat en informatique de l'Université de Caen. 29 novembre 1996 |
| Un contexte applicatif : la bureautique communicante ; un environnement technologique : les systèmes répartis à objets ; un cadre théorique : les systèmes multi-agents.
La répartition et le besoin croissant d'ouverture du système d'information de l'entreprise rendent déterminantes les capacités d'autonomie et de coopération de ses composants. Dans cette optique, la technologie des systèmes répartis orientés objets se montre particulièrement pertinente, par son support immédiat à la répartition fonctionnelle et géographique des problèmes. Mais autonomie et coopération nécessitent également la mise en oeuvre de modèles et d'outils de haut niveau pour permettre raisonnement, négociation, et adaptation dans un système réparti, ouvert et dynamique. C'est ainsi que les recherches en Intelligence Artificielle Distribuée s'intégrent aux systèmes répartis à objets. Le lien entre ces deux domaines est double : d'une part, toute application répartie orientée objets peut être vue comme un système multi-agents et tirer parti des modèles de coopération, de comportement et de représentation des connaissances développés en IAD. Réciproquement, les systèmes répartis à objets apportent une technologie adaptée à la réalisation pratique, la simulation et l'observation de systèmes multi-agents. Conformément à ces principes, nous proposons une représentation multi-agents des systèmes de bureautique communicante, et un environnement d'implémentation d'agents à tendance cognitive. Basé sur l'intégration d'un interpréteur Prolog dans les objets C++ du système réparti à objets COOLv1, il est utilisé pour illustrer l'approche multi-agents dans le cadre d'une maquette d'application. Enfin, dans une problématique de mise en relation optimale d'un agent client avec un agent serveur, nous soulignons la nécessité de structures et protocoles de coopération/négociation mettant en oeuvre des techniques d'Intelligence Artificielle. |
|
| Supporting intelligent agents in a distributed environment: a COOL-based approach | |
| référence : | Technology of Object-Oriented Languages and Systems. Versailles, 1995. TOOLS 16, Prentice Hall, ISBN 0-13-443128-6, p. 235-246. |
| co-auteur : | François Bourdon |
| The background of this article consists in a distributed object oriented approach for the development of logically and physically distributed applications in the field of Computer Supported Cooperative Work. The use of an object oriented distributed system (COOL) is a first step towards the implementation of solutions to these naturally distributed problems. But this support is insufficient to directly implement high-level interactions and behaviour: the entities have to adjust themselves to a dynamic system where services and servers appear, evolve, and disappear.
This article introduces both a model and a tool which are designed to enhance the entities cooperation ability and autonomous adaptability. The model springs from investigations in the field of Multi-Agent Systems and consists in a uniform representation of every resource of the system (hardware, software, user) by agents. The implementation tool and associated developments are described in some detail. They are based on an enhanced Prolog interpreter integrating the main COOL features: various communication modes, active objects, migrating objects... Our Prolog-based approach is discussed and illustrated by the implementation of a distributed application and some generic multi-agent system structures and protocols for cooperation. |
|
| Towards a multi-agent model for the office information system: a Prolog based approach | |
| référence : | 3rd International Conference on the Practical Application of Prolog. Paris, avril 1995. ISBN 0 9525554 0 9, p. 191-200. |
| co-auteur : | François Bourdon |
| The background of this article consists in a distributed object oriented approach for the development of logically and physically distributed applications in the field of Computer Supported Cooperative Work. The use of an object oriented distributed system (COOL) is a first step towards the implementation of solutions to these naturally distributed problems. But this support is insufficient to directly implement high-level interactions and behaviour: the entities have to adjust themselves to a dynamic system where services and servers appear, evolve, and disappear.
This article introduces both a model and a tool which are designed to enhance the entities cooperation ability and autonomous adaptability. The model springs from investigations in the field of Multi-Agent Systems and consists in a uniform representation of every resource of the system (hardware, software, user) by agents. The implementation tool and associated goodies are described in some detail. They are based on an enhanced Prolog interpreter essentially integrating the COOL features: various communication modes, active objects, migrating objects... Our Prolog-based approach is discussed and illustrated by the implementation of a distributed application and some generic multi-agent system structures and protocols. We also show how several artificial intelligence techniques are involved in the solutions we propose. |
|
| Une approche Multi-Agents des systèmes de bureautique communicante | |
| référence : | Journée "Systèmes Multi-Agents" du PRC-GDR Intelligence Articielle. Paris, 16 décembre 1994. Yves Demazeau, Anne Collinot éditeurs. |
| co-auteur : | François Bourdon |
| Après avoir esquissé les besoins en répartition et en autonomie des systèmes d'information, cet article se propose de décrire une modélisation multi-agents des systèmes de bureautique communicante et une implémentation dans le cadre d'une application concrète. Cette mise en oeuvre est réalisée avec PUMA, une plate-forme basée sur l'intégration d'un interpréteur Prolog à l'environnement réparti à objets COOL, que nous présentons comme un outil d'IAD. Nous concluons en discutant des ouvertures apportées par le modèle multi-agents et le principe de l'outil. | |
