Bibliothèque de L'IMIST/CNRST

PH.D Université Mohammed V 2017

Network selection receives considerable attention from communication networks researchers. This remarkable interest is motivated by the variety of the existing technologies and the offered services. The widespread use of these services leads to growth in the expectations and requirements of mobile users in terms of efficiency.
This thesis focuses on proposing and evaluating new network selection models for heterogeneous wireless networks, and providing a real-time selection of always best connected network while maintaining the QoS for different multimedia services. The main contribution consists of the proposal of a model of multi-criteria network selection. This model is open in the sense that it can accommodate the implementation of different parameters in the decision of the network that may be selected, considering the type of application used by the mobile user. This thesis is developed from two aspects: Network selection algorithms proposition and performance study.
In the first part, three complementary solutions were proposed to enable a fair and thoughtful allocation of resources and serve a maximum number of users.
First, we study the problem of network selection using the decision-making strategy MADM (Multi-Criteria Decision Making). It consists of choosing the best network according to several criteria using a batch of weights generated with the AHP method. We propose, therefore, a weight improvement based on fuzzy logic, FAHP. In order to take into account the speed of mobile users, we propose a context-aware network selection based on a new utility function that takes into account the user's preferences and the quality of service requirements in a high mobility scenario. We also improve the latter solution by proposing a network selection algorithm that provides a load balancing technique based on matching game theory to overcome network overload conditions in high-speed scenarios.
In the second part, the performance study has been evaluated for two scenario types: Cellular and Vehicular. The proposed algorithms were evaluated and validated in mobile and vehicular environments using the NS3 simulator.