Home > Research > Publications & Outputs > Scalable and Efficient Multiagent Platform Clos...
View graph of relations

Scalable and Efficient Multiagent Platform Closer to the Operating System

Research output: Contribution in Book/Report/ProceedingsConference contribution

Published
  • Juan M. Alberola
  • Jose M. Such
  • Agustin Espinosa
  • Vicent Botti
  • Ana Garcia-Fornes
Close
Publication date2008
Host publicationArtificial intelligence research and development: proceedings of the 11th International Conference of the Catalan Association for Artificial Intelligence
EditorsTeresa Alsinet, Josep PuyolGruart, Carme Torras
Place of PublicationAsmsterdam
PublisherIOS Press
Pages7-15
Number of pages9
ISBN (Print)978-1-58603-925-7
<mark>Original language</mark>English
Event11th International Conference of the Catalan-Association-for-Artificial-Intelligence - St Marti Empuries

Conference

Conference11th International Conference of the Catalan-Association-for-Artificial-Intelligence
CitySt Marti Empuries
Period22/10/0824/10/08

Conference

Conference11th International Conference of the Catalan-Association-for-Artificial-Intelligence
CitySt Marti Empuries
Period22/10/0824/10/08

Abstract

Agent technology demands Support to build complex. open and large-scale applications. Therefore, efficient, scalable and robust Multiagent Platforms are needed. The performance of current Multiagent platforms, which are usually developed in Java, has been tested in different studies. Results presented in these studies show that the performance of these Multiagent Platforms is not as good as one could expect. We present in this paper a different approach of a Multiagent Platform closer to the Operating System level to improve the performance achieved. This platform is focused on offering the main services required by a Multiagent platform and also improving the functionality of some of them, such as the messaging service. In this sense, we extend this functionality with advanced communication mechanisms such as agent groups and predefined interaction protocols. We also present a performance evaluation of the current platform implementation.