Découverte de services web via le Cloud computing à base d’agents mobiles

Abstract

With the rapid evolution of cloud services, in number and functionality, the need for a discovery and service composition engine becomes an unavoidable necessity. To achieve such a system, we need to use appropriate technologies to the nature of this type of application, namely, agents and cloud computing technologies. The agent technology is very suitable for modeling systems that require collaboration between several entities in order to achieve a common goal, such as performing a system where agents work together to find the best web services that meet the functional and non-functional client requirements. Moreover, the distributed nature of cloud services requires the use of mobile agents to search and create secure communications between the search engine and the web service providers and to reduce costs. The cloud computing technology has a significant advantage for the distributed Internet applications. On the one hand, this technology allows clients to pay for virtual and software resources that they use and avoid buying the new versions of the cloud system that they use. On the other hand, this technology enables using physical and virtual ressources with unlimited storage and treatment capacities. In our work we are interested to study and propose a discovery and composition engine which has the capacities to treate thousands of queries and WSDL files simultaneously with the best response time. In this thesis we propose a new Cloud computing architecture based-agent for SaaS services discovery and composition. This architecture is intended to support and implement a new selection, composition and ranking mechanism. This mechanism is based on a new measure, so-called "existence degree", this measure allows us to calculate the degree of adaptation between the client query and the tree representations of WSDL files. In addition, the discevery and composition mechanisms combine the functional, QoS and SaaS services parameters to create composite services that meet the functional and non functional client requirements. The simulation results show the effectiveness of the discovery and composition engine especially for response time. The focus is on the need to develop an economic model which is adapted to the operations and techniques of this kind of engine. Finally, we proposed a comparison and evaluation model in order to compare the proposed approach with the other approaches from the same domain