Thèse soutenue

Auteur / Autrice : Abbas Antoun Hatoum
Direction : Guy Pujolle
Type : Thèse de doctorat
Discipline(s) : Informatique
Date : Soutenance en 2013
Etablissement(s) : Paris 6


FR  |  

Recently, operators have resorted to femtocell networks in order to enhance indoor coverage, network capacity and quality of service since macro-antennas alone fail to reach these objectives. Nevertheless, they are confronted to many challenges. To successfully deploy such solution, efficient resource allocation algorithms and interference mitigation techniques should be deployed. In this thesis, we address the issue of resources allocation in femtocell networks using OFDMA technology (e. G. , WiMAX, LTE). Specifically, we first propose a hybrid centralized/distributed resource allocation strategy for split spectrum namely Femtocell Cluster-based Resource Allocation (FCRA). Firstly, FCRA builds disjoint femtocell clusters. Then, within a cluster the optimal resource allocation for each femtocell is performed by its clusterhead. Finally, the contingent collisions among different clusters are fixed. To achieve this, we formulate the problem mathematically as Min-Max optimization problem. Then, a co-channel resource allocation algorithm (CO-FCRA) introduces spectrum sharing between femto and macro users. Spectrum sensing approaches are used to detect existing neighboring transmissions in the uplink and estimates resources used in the downlink to allocate resources accordingly. In a second approach, we consider networks with quality of service differentiation among users and propose a new algorithm, namely (Q-FCRA) with both high priority and best effort users. The optimization problem is modified to take into account both user types and allocates resources accordingly. The objective is to maximize the number of accepted high priority users and allocate as much as possible best effort users. As a third contribution, we present a power control algorithm (QP-FCRA), where femto stations allocate both resource blocks and transmission power on the different channels to effectively mitigate interference within the same cluster and increase the spectrum spatial reuse. The transmission power is calculated based on the interference received to satisfy a minimum required SINR threshold. Several existing works have been used for comparison. Different network densities, interference levels, session duration and mobility rates have been considered. Performance evaluation shows the improvement and the outperformance of our algorithms compared to the existing techniques regarding different performance metrics such as the number of accepted and rejected users, the fairness, the throughput satisfaction rate, the spectrum spatial reuse and the convergence and computation time. The scalability of our algorithm compared to the centralized ones is proven as well as the performance compared to the distributed algorithms.