Gestion des ressources et de la qualite de service dans un reseau mobile heterogene de prochaine generation

Since there is no single mobile wireless technology that can provide both high system capacity and cost effective global service coverage for more demanding mobile users, existent and future heterogeneous mobile networks have to be efficiently integrated for this purpose profiting to an increasing number of multi-mode mobile stations. This integration efficiency in what we call a next generation heterogeneous mobile network (NGHMN) can be achieved using several approaches that we investigate in this thesis.;In a first contribution, we provide models and algorithms in a newly proposed generic analytical framework for accurately evaluating call level quality of service (QoS) parameters of any kind of NGHMN implementing vertical handoff and different network selection strategies. Using this analytical framework, we estimate and compare the handoff rates and the blocking probabilities under a standard Exponential model, a general Gamma model and an accurate Hyper-exponential model of cell residence times in 3G cells and WLAN cells. Results show that given the different variabilities of the cell residence times, blocking probabilities under the Exponential model are over-estimated for 3G cells and largely under-estimated for WLAN cells. Besides these blocking probabilities are over-estimated under the Gamma model for WLAN cells.;In a second contribution of this thesis, we propose, analyze and validate an optimal voice admission control algorithm in a 3G/WLAN integrated network. We provide analytical models, validated by extensive computer simulations, for evaluating optimal admission control blocking performance under different wireless bandwidth allocation schemes for vertical handoff calls and under redefined dropping probabilities taking in account different coupling schemes. Performance results show that, compared to the tight coupling bandwidth allocation scheme, the loose coupling one provides a good mean to limit the increase of blocking probabilities in expensive and high-loaded 3G cells when integrating both 3G and WLAN network. The proposed models can help network designers figure out the performance of an eventual type of integration between 3G and WLAN.;In a last contribution of this thesis, we evaluate the impact of mobility and soft handoff on the performance of optimal voice admission control in loosely coupled 3G/WLAN networks. For this purpose, we propose accurate mobility and wireless bandwidth allocation models tacking in account the soft vertical handoff in this kind of NGHMN and we analyze the blocking probabilities of optimal voice admission control under different soft vertical handoff algorithms. Using these models, we show that an optimal resource-efficient soft handoff algorithm (RESHO) performs significantly better than a standard static-threshold soft handoff algorithm (STSHO) particularly in WLAN mobility environments. In fact, results show that 3G new call blocking probability reduction gained by using RESHO compared to STSHO is largely increased when mobile station (MS) velocities have low mean and high variability which typically characterizes the WLAN mobility environment. Besides, we show that RESHO reduces all blocking and dropping probabilities. We believe that the provided models and the presented results could help design efficient MS controlled soft vertical handoff algorithms for emergent loosely coupled 3G/WLAN networks.