BERTIER Clément

PhD student at Sorbonne University
Team : NPA
https://lip6.fr/Clement.Bertier

Supervision : Marcelo DIAS DE AMORIM

Co-supervision : BENBADIS Farid, CONAN Vania

Quantification in Device-to-Device Networks: from Link Estimation to Graph Utility

Device-to-device (D2D) communications are valuable in several domains, such as data offloading and diffusion, as their cost is only a fraction of what regular cellular communication would have. In this thesis, we argue that understanding the potential utility behind direct communications is key to quantifying the realization of contact networks.
Firstly, we consider the problem of estimating the importance of a node in large dynamic topologies. We propose a novel approach to estimate centralities based on a pre-established database, where the estimation is based on the geographical coordinates of the node instead of the identifier of the node. Doing so enables us to estimate the centrality of a node for a fraction of the computational cost. Secondly, we quantify the value of direct links through an experimental measurement campaign. Using an Android tool of our making, we derived a model to obtain an estimate of the upper-bound of D2D throughput based on the distance between the devices. Thirdly, we investigate the differences between the traditional quantification of a contact and the model extracted from our measurements campaigns. Among other results, we reveal that when considering an adaptive throughput according to the distance between two devices, the long-distance data-exchange makes up more than 50% of the total data exchanged in the entire network. We propose a tool to extract from mobility datasets the volume of data obtained, based on specific contact quantification strategies.

Defence : 01/13/2020

Jury members :

Katia Jaffres-Runser, Maître de conférences, INPT-ENSEEIHT [Rapporteur]
Fabrice Valois, Professeur, INSA Lyon [Rapporteur]
Aline Carneiro Viana, Maître de conférences, INRIA Saclay
Pierre Sens, Professeur, Sorbonne Université
Farid Benbadis, Ingénieur de recherche, Thales SIX GTS
Vania Conan, Ingénieur de recherche, Thales SIX GTS
Marcelo Dias de Amorim, Directeur de recherche, CNRS/Sorbonne Université

Departure date : 01/13/2020

2018-2020 Publications