Adaptation de services dans un espace intelligent sensible au contexte

Abstract

With the appearance of the paradigms of ambient intelligence, new ambient intelligent systems are emerging with the aim to create and manage intelligent environments. These environments are smart spaces characterized mainly by the openness, the heterogeneity, the uncertainty and the dynamic of the entities that constitute them. These characteristics involve significant scientific challenges for the design and implementation of an appropriate intelligent system. These challenges are mostly abstraction and management of context, reactivity to events detection, context-awareness and self-adaptation to unpredictable changes that may occur in an ambient environment. In this thesis, we proposed an architecture for system capable of adapting services according to user needs by taking into account, on the one hand, the context of the ambient environment and his various equipment and, on the other hand, the varying needs expressed by users. This system is built according a context aware model, which is adaptive and reactive to events detection. It is based on modular entities with strong cohesion and low coupling to be flexible and effective. This system also includes a module to adapt services in order to identify context and adjust dynamically the selected service according to the user requirements. This adaptation is completed by two algorithms. The first one is an reinforcement algorithm (Qlearning), and the second is an supervised algorithm (CBR). The intention of this hybridization is overcome the drawbacks of Q-learning algorithm to highlight a new approach able to handle context, select and adapt service.

Date	23 Oct 2018
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Chakib Tadj (Supervisor) & Miraoui Moeiz (Co-supervisor)