Franck Legendre - Mobility Modeling for Self-Organizing Networks

Date: February 2nd 2007, 2:30-4:30 pm

Room: 549

Abstract

Research on mobility modeling has reached a critical point. In spite of the large number of available proposals, networking evaluation is still done with acknowledged models such as random waypoint or reference point group mobility. We believe it is time to federate mobility models under a common formalism that allows specifying, in a simple and common fashion, a broad range of mobility models. We propose Ghost, a meta-modeling approach to design mobility models. The core of Ghost is composed of (i) atomic units governing mobility namely behavioral rules, (ii) an activity-task formalism, and (iii) a fluid approach. Ghost enables also to calibrate and validate models with empirical measurements in order to reproduce the mobility characteristics observed in reality. These approaches and functionalities lead Ghost to a higher complexity than classical mobility simulators; but it also lead to the generation of more realistic traces. Ghost is also interactive. It is associated with a tool through which users can define scenarios using a high-level language and interact with on-going simulations. Ghost can also be bound to a network simulator to run joint simulations. We illustrate the power of Ghost by investigating its ability to easily represent existing mobility models but also new pedestrians models of indoor (corridor and storey building) and outdoor mobility (campus).

Bio

Dr Franck Legendre is currently a postdoctoral fellow at the Networks and Performances Analysis group of the LIP6 laboratory, University Pierre et Marie Curie~-- Paris VI. He received a Computer Science Ph.D. from University of Paris VI in 2006, M.Sc. degrees in Telecommunications Engineering from Institut National des Telecommunications (INT), Evry, France, and in Computer Networks from University of Paris VI in 2002. His research interests focuses on mobility models and routing considerations for the support of self-organizing network composition.