Offre de Thèse CIFRE - Huawei / LIMOS : Network optimization with non-linear performance models

Title: Network optimization with non-linear performance models

Supervisors: Pr. Mourad BAIOU, Dr. Hervé KERIVIN and Dr. Sébastien MARTIN

Keywords: optimization, telecommunication networks, mathematical programming, exact algorithms, non-linear model, data driven optimization.

Abstract: Quality of Service (QoS) and routing policies are key to control how bandwidth is shared among different applications in computer networks. They can be tuned to satisfy strict Service Level Agreements (SLA) of applications in terms of end-to-end delay, packet loss and jitter. They can also be used to meet global objectives in terms of minimizing financial expenses or maximizing the experienced network quality. In this context, conflicting objectives may arise when optimizing routing and QoS policies.  This thesis will focus on the integration of QoS in the classical multi-commodity flow problem. The QoS can be modeled, for instance, with non-linear function thanks to the network calculus. The primary goal will be to develop a mathematical optimization tool driven by the QoS requirements. Based on this research direction, some theoretical results will be proposed to enforce the efficiency of exact method or to prove the efficiency of some derived heuristics. Mathematical optimization will be based on branch-and-cut and branch-and-price algorithms to solve with exact method. Some piece-wise technics of non-linear function can be applied to provides some bounds. The novelty of the thesis will be in the integration of non-linear functions to drive the mathematical optimization for considering the QoS.

The thesis will be developed in the Mathematical and Algorithmic Sciences Lab, Paris Research Center of Huawei and in CNRS laboratory LIMOS at University Clermont Auvergne. This work will be done in collaboration with prof. Walid BEN-AMEUR of Télécom SudParis. The focus is on the development of scientific results concerning theory and methodology which may be of interest for the mathematical optimization community. A special attention will be given to the publication of the results in top journals of the domain (like Mathematical Programming, Operation Research as well as top telecommunication conferences like INFOCOM).

The PhD grant will be funded by Huawei France Research Center (FRC) located in Boulogne-Billancourt, Paris area. The center is leading advanced research in the fields of Algorithm and Software design, Aesthetics, MBB & Home devices and Parallel Computing, to create and design the innovative technologies and software platforms for the company. The project is in collaboration with University Clermont Auvergne, LIMOS (https://limos.fr/). LIMOS is a well-known research center in optimization. The PhD student will perform his research activities between LIMOS and the Paris research center of Huawei.

Skills: The PhD candidate is expected to perform scientific research in the domain of network optimization for telecommunications. Applicants should demonstrate good programming skills and deep knowledge in solution methods for combinatorial optimization problems. Experience with branch-and-cut algorithms and/or branch-and-price algorithms is considered as plus.  Applicants should have completed (or close to completion of) a master degree in operations research, computer science, mathematics or a comparable domain. Fluency English is required. Moreover, the candidate must be comfortable in programming in C/C++. Background knowledge on solution methods to handle non-linear problems is considered as plus. Affinity with industrial Engineering or telecommunication networks is considered as plus. 

Application: must be submitted electronically to herve.kerivin@isima.fr, walid.benameur@telecom-sudparis.eu and sebastien.martin@huawei.com and must include a cover letter, a curriculum vitae detailing educational background research experience, up to three recommendation letters, copy of master thesis and copies of undergraduate and graduate transcripts. The position is for 3 years starting as soon as possible. 

Deadline: application must be submitted as soon as possible. We will accept later applications if the position is not yet filled.