Title: How to select dynamically a QoS-driven composite web service by a multi-agent system using CBR method

Authors: Fatma Siala; Khaled Ghedira

Addresses: Department of Computer Sciences, University of Tunis, SOIE - Higher Institute of Management (ISG), 41, Liberty Street, Bouchoucha 2000 Bardo, Tunis, Tunisia ' Department of Computer Sciences, University of Tunis, SOIE - Higher Institute of Management (ISG), 41, Liberty Street, Bouchoucha 2000 Bardo, Tunis, Tunisia

Abstract: Service-oriented architecture permits the composition of web services provided with different Quality of Service (QoS) levels. In a given composition, finding the set of services that optimises some QoS attributes under its constraints is a problem that needs to be solved. Our aim is to propose an intelligent approach to the selection of a Composite Web Service (CWS) based on QoS. This paper reports the authors' recent research on addressing the issue. An overview on the previously proposed approaches is presented. These approaches correspond to several improvements of an existing multi-agent one, which is well-cited in the specialised literature. Each framework, implemented on JADE (Java Agent Development) framework, improves another one in terms of CPU time and/or QoS score, to reach a new agent-based and scalable framework. The last framework utilises the agents' ability of negotiation, interaction and cooperation in order to facilitate the selection of composite web services. By using CBR method, the agents can memorise QoS scores and availability. The improvements are related not only to the CPU time but also to the Composite QoS (CQoS) value, while operating in a dynamic environment and taking into account user preferences.

Keywords: composite web services; QoS; quality of service; multi-agent systems; MAS; negotiation; availability; CBR; case-based reasoning; agent-based systems; interaction; cooperation.

DOI: 10.1504/IJWMC.2014.063054

International Journal of Wireless and Mobile Computing, 2014 Vol.7 No.4, pp.327 - 347

Available online: 27 Jun 2014 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article