PhD graduated
Team : Phare
Departure date : 11/30/2011

Supervision : Guy PUJOLLE

Co-supervision : LANGAR Rami

An Autonomic Architecture for Wireless Networks : Proposition and Evaluation Methodology

Autonomic network management is a promising approach to reduce the cost and the complexity of managing network infrastructures. It relates to the capability of the network to fulfill self-configuration, self-optimization, self-protection and self-healing functionalities. To achieve these properties, a new network management solution based on autonomic architecture is paramount. The objective of this thesis consists in providing an autonomic architecture for wireless mobile networks which is able to self-adapt according to their dynamic context. To achieve this, we proposed SADA, a self-adaptive autonomic architecture for wireless mobile networks. SADA presents two new autonomic features: first, it enriches the IBM reference autonomic model exploited by existing architectures with self-adaptation capabilities. Second, it defines a second autonomic control loop around the monitoring component, allowing it to self-adapt according to the network context. The proposed SADA architecture provides the self-adaptation features based on a learning mechanism which uses the experienced outcome of its decisions to enhance its operation. Moreover, the SADA architecture uses a self-adaptive monitoring approach which provides required information for adaptation algorithms with a minimum extra overhead. As a case of application, SADA has been employed in routing service, resulting in the proposal of a self-adaptive routing scheme. Simulation results showed that SADA improves significantly the network performance in terms of packet delivery ratio, average end-to-end delay and routing overhead compared to the classical IP-based architecture. Moreover, we applied the SADA architecture to design a self-adaptive knowledge monitoring scheme for autonomic network trust management. The objective of this case study was to investigate the effectiveness of the autonomic monitoring mechanism proposed by the SADA architecture. Lastly, we proposed a quantitative evaluation methodology to evaluate and compare the efficiency of autonomic architectures from the autonomicity standpoint. To show the applicability of our methodology, we considered two use cases in two different contexts: (i) a wireless ad-hoc network using SADA architecture, and (ii) an infrastructure-based wireless network. Results showed that the proposed SADA architecture constitutes an efficient autonomic solution, characterized by its distributed intelligence, learning capability, self-adaptation features, low complexity, light monitoring, high quality of knowledge, security support and evolvability.

Defence : 11/25/2011 - 10h45 - Site Jussieu 25-26/105

Jury members :

Francine Krief, Professeur à l'ENSEIRIB [Rapporteur]
Otto Carlos Muniz Bandeira Duarte, Professeur au Brésil [Rapporteur]
Yacine Ghamri-Doudane, HDR à l'ENSIIE
Marcelo DIAS DE AMORIM, Directeur de recherche CNRS, UPMC
Guy Pujolle, Professeur à l'UPMC
Rami Langar, Maître de conférence à l'UPMC

2008-2018 Publications