A Dual Simulation Environment for Simulating MAS in Telecommunication Networks NADA MESKAOUI DOMINIQUE GAITI KARIM Y. KABALAN Lebanese University University of Troyes American University of Beirut Hadath – Lebanon 12 Rue Marie Curies Bliss Street Lebanon Troyes, Cedex Beirut - Lebanon nadames@inco.com.lb dominique.Gaiti@utt.fr kabalan@aub.edu.lb Abstract— this paper presents a dual simulation environment for simulating different type of telecommunication networks integrating intelligent agents. Agents are considered as advanced tools for resolving complex issues in networking based on intelligent and dynamic features. To test the efficiency of these proposals, new simulation environments, integrating both agents and network components, are required. In this paper we propose an extension with intelligent capabilities to a networking platform. This dual simulation environment has been tested for implementing agents in a DiffServ network to improve its performance. Simulation results show the efficiency of integrating agents within telecommunication networks and also prove that such a dual simulation environment is needed to test new techniques based on agents and multi-agent systems in networking. Key-Words: Artificial intelligence, telecommunication, networks, single-agent systems, multi-agent systems, Diffserv network, platforms. 1 Introduction New approaches, in any domain, have to be validated by simulations before its real implementation. For this reason, different types of platforms were conceived in various domains to help researchers testing the efficiency of their proposals. Some Existing platforms are dedicated to simulate telecommunication networks and others for multi-agent systems. Networking platforms do not have intelligent capabilities. Also, platforms conceived for multi-agent systems’ simulation stress on the definition of the intelligent capabilities of the agent without taking into consideration the telecommunication networks’ elements. Network elements could be modeled and built within these platforms but this means that every networking component should be redefined, modeled and represented by agents [1]. Standard components and mechanisms that constitute a telecommunication network are not provided in this type of platforms. The objective of our work is to propose an intelligent framework to model and simulate intelligent features in networking in order to prove the efficiency of intelligent approaches – based on agents and multi-agents systems – in the improvement of telecommunication networks’ performance. This paper presents a platform having the ability to simulate both networks and multi- agent systems. Having this platform, one can specify any type of agent’s behavior to interact with any node structure. The proposed platform is detailed in section 4 while sections 2 and 3 present, respectively, the existing platforms for networks and Multi-Agent systems simulations. Section 5 shows some simulation results and a conclusion is provided in section 6. 2 Platforms for network simulation In telecommunication networks simulations, two platforms’ categories are identified: open platforms for standard network simulations able to implement new techniques and network WSEAS TRANSACTIONS on COMPUTERS Nada Meskaoui, Dominique Gaiti, Karim Y. Kabalan ISSN: 1109-2750 916 Issue 6, Volume 8, June 2009