Soutenance de thèse de doctorat le 19 Décembre 2019, 15h00 à CentraleSupelec (Gif-sur-Yvette) Amphi F3-06
|M. Marco Di Renzo
|Directeur de recherche-CNRS
|Directeur de Thèse
|Mme Maryline Helard
|M. Jalel Ben-Othman
|Professeur-CNRS-CentraleSupélec-Université Paris-Saclay and Université Paris 13
|M. Jean-Marie Gorce
|Mme Lina Mroueh
|Maître de conférences-ISEP
|Mme Valeria Loscri
|Chargé de recherche-Inria Lille-Nord Europe
|M. Mustapha Benjillali
|Maître de conférences-INPT-Maroc
|M. Laurent Clavier
Abstract: In this thesis, we have developed new analytical frameworks for analyzing and optimizing future cellular networks with the aid of stochastic geometry and point processes. This thesis provides four main technical contributions.
First, we analyze emerging networks that can communicate by using light instead of radio waves. In this context, we propose an innovative analytical framework that allows us to estimate the coverage probability and the average rate of spatially distributed networks, which are used to gain insight for system optimization.
Second, we propose an innovative methodology for modeling spatially correlated cellular networks by using in-homogeneous point processes. The proposed approach is tested against practical deployment of cellular networks and found to be tractable and accurate. It is applied to the analysis of visible light communication networks, and the impact of spatial correlation is studied.
Third, we tackle the open problem of modeling Massive MIMO cellular networks. We study uplink and downlink cellular networks and propose new upper and lower bounds for the average spectral efficiency, which allow us to identify the optimal number of user to serve in each cell of the network and the impact of several key system parameters.
Fourth, we introduce and analyze the performance of a new interference-aware scheduling algorithm for application to the uplink of cellular networks. The proposed approach is based on muting some users in order to reduce the level of interference. The achievable performance and the user-fairness of the proposed approach are discussed and quantified analytically.
This PhD thesis is supported by the European Commission through the H2020-ETN-5Gaura project under grant 675806