Calendrier des animations scientifiques

TBA

Mes travaux portent sur les réseaux à longue portée et faible consommation énergétique basés sur la technologie LoRa. Je m’intéresse dans un premier temps à la capacité des réseaux LoRaWAN en terme de nombre de nœuds tout en garantissant un taux de livraison des transmissions. J’optimise l’allocation des facteurs d’étalement des noeuds permettant d’assurer des transmissions fiables en fonction de la distance aux passerelles, à partir d’un modèle de réseau réaliste tout en respectant les contraintes règlementaires d’utilisation de la bande de fréquence. J’étudie également l’ordonnancement des transmissions montantes et descendantes afin d’améliorer les protocoles en pratique. Les résultats montrent l’intérêt du contrôle de puissance et de la répétition des envois de paquets et ouvrent des perspectives sur l’utilisation de noeuds relais dans ce type de réseau.

Online bin stretching is an online packing problem where some of the best known lower and upper bounds were found through computational searches.

In this problem a finite sequence of items, each defined by a real number denoting its size, must be placed into m bins. Items are revealed sequentially: whenever an item is revealed, it must be placed irremediably into a bin before the next item is shown. The whole set of items given is also known to fit inside m bins of unit size. The objective is to place all the items into the bins to minimize the load of the fullest bin. This is similar to the online scheduling problem on m machines with known makespan: Pm | Online, Cmax = 1 | Cmax.

We consider computational searches that give verifiable trees which prove lower or upper bounds on the best possible online algorithms. The limiting factor in obtaining better bounds with such methods is the computational time allowed.

We give a theoretical guarantee that such methods do converge towards the optimal online performance. These results frame a theoretical framework for showing the convergence of computational approaches for online problems.

In collaboration with Nicolas Catusse and Nadia Brauner.
Link to Arxiv preprint: https://arxiv.org/abs/2506.17271

TBA

Decision-Dependent Information Discovery (DDID), a variant of two-stage robust optimization, addresses problems where uncertainty can be reduced by querying the values of some uncertain parameters before making decisions for the nominal problem. In this work, we study DDID applied to combinatorial problems under a robust capacity constraint. In our setting, the number of uncertain parameters that can be queried in the first stage is limited by a given integer q. We show that if q is considered as a constant, then some classical nominal combinatorial problems—such as selection problem, spanning tree problem and assignment problem—remain polynomial-time solvable in this robust optimization setting. Specifically, we prove that their objective functions can be computed by solving polynomial number of (compact) linear programs.