DIAS DE AMORIM Marcelo

Habilitation
Team : NPA
Departure date : 10/21/2008
https://lip6.fr/Marcelo.Amorim

Mobile multi-hop wireless networks: Conceptual and practical concerns

This HDR is about multi-hop wireless networks; more specifically about routing and (some) related issues. Routing is a general term that finds applications in numerous areas such as networking, transportation systems, electric circuits, postal systems, and neurobiology, to cite a few. The class of networks we address is the so-called wireless self-organizing networks (SON). Loosely speaking, a self-organizing network lacks a priori infrastructure that would be present in a traditional network at power-up time. This lack of a priori infrastructure and the need for (sometimes frequent) infrastructure updates pose design requirements that are not present in traditional networks. The set of contributions in this HDR is classified into two types: conceptual and support. In the “conceptual” part, we cover the routing building blocks with the following contributions. We start by considering the problems of addressing and locating and present an approach, namely indirect routing, that borrows some concepts of peer-to-peer overlays and applies them at the network layer. Our first contribution is Tribe, an algorithm that organizes nodes in a tree and that has the advantage of providing an inherent forwarding structure. Managing mobility and churn with a tree is however difficult; as a response to this issue, we propose Twins, an addressing and location scheme that uses Hilbert space filling curves to keep the simplicity of one-dimensional location spaces while relying on a multi-dimensional data forwarding space. Our second set of contributions concerns positioning (i.e., assigning coordinates to nodes). We focus on the class of virtual positioning where nodes are assigned coordinates in a relative fashion. Our main outcome is Jumps, an algorithm that relaxes the constraint of three dimensions imposed by previous approaches. Finally, we address the problem of location dissemination. We consider a new class of routing protocols that relies on exponential discovery systems. The traditional approach relies on encounters between nodes to disseminate location information. Since mobility is the keyword of such a scheme, we propose ELIP, an original algorithm that also takes advantage of the mobility of data packets. The second part of this HDR focuses on more practical issues of routing on self-organizing networks. The first issue is related to mobility modeling. We contribute to this domain by modeling the causes of mobility instead of its consequences. We propose a new class of mobility models relying on the concept of behavioral rules. Our framework, namely Ghost, provides a methodology to combine atomic forces to generate acceleration on an individual basis. We rely on this methodology to propose SIMPS, a social mobility model that generates mobility patterns which are coherent with recent empirical results. The third contribution is related to the evaluation of protocol and algorithms under real conditions. We propose Prawn, a software environment for rapid prototyping of protocols and algorithms on top of nodes equipped with IEEE 802.11 interfaces. With similar goals in mind, we are developing a passive measurement system, namely WiPal, to measure the mobility of nodes in open environments. The idea is to gather as much information as possible and feed Otiy, a new user-centric location service for wireless mesh networks.

Defence : 10/21/2008

Jury members :

Apporteurs : Anne-Marie Kermarrec, DR INRIA Rennes
Sergio Palazzo, Prof. Università degli Studi di Catania
Eric Fleury, Prof. École Normale Supérieure de Lyon
Examinateurs : Olivier Festor, DR INRIA Nancy – Grand-Est
Jean-Jacques Pansiot, Prof. Université de Strasbourg
Yannis Viniotis, Prof. North Carolina State University
Serge Fdida, Prof. UPMC Univ Paris 06

Director of research

17 PhD graduated 2006 - 2023

Mentions légales
Site map