Members

Jonathan LEDY

Research Engeneer

Contact

Photo ledy Universite de Haute Alsace
IRIMAS
Ecole Nationale Superieure d Ingenieurs Sud Alsace (ENSISA)
12 rue des Freres Lumiere, 68093 MULHOUSE Cedex, France
Phone: +33 (0)3 89 33 69 54
Fax: +33 (0)3 89 42 32 82
Courriel : jonathan.ledy(at)uha.fr

Research Topics

Vehicular Adhoc NETworks simulations

Propagation models
Routing protocols
Cross-layering
Link-failure prediction

PhD Thesis

Direction : Rodolphe VAUZELLE, Benoit HILT, Herve BOEGLEN et Anne-Marie POUSSARD
Financement : Bourse Regionale
Periode : 10/2009-10/2012
Soutenance : 19/12/2012

PhD Title

Strategie d adaptation de liens sur canaux radios dynamiques pour les communications entre vehicules - Optimisation de la Qualite de Service

Abstract

This thesis deals with the optimization of communications in vehicular networks by using a realistic simulation platform. A realistic environment implies the usage of mobility models adapted to vehicles and also highly detailed physical models (channel models and digital transmission chain). The first part of our work has consisted in the design of a realistic simulation platform dedicated to VANETs (Vehicular Ad hoc NETworks). This platform has been completed by a semi-deterministic propagation model which we have designed. This model called UM-CRT has the advantage to have the same level of realism than a deterministic model while requiring much less computation time. This model has been validated by comparison with a deterministic ray tracing simulator. We then have used this platform to evaluate routing protocols. The efficiency of different ad hoc routing protocols in realistic conditions has led us to focus our study on the family of reactive protocols. From this evaluation we have selected AODV (Ad hoc On demand Distance Vector) to which we have applied a cross-layer metric in order to reduce the performance degradation caused by the realistic environment. We then have used a tuning technique with reactive protocols. Finally, we have evaluated several SISO and MIMO physical layers. This work shows that only improvements combined at different levels (physical and network) can yield a significant increase in performance.