BELBLIDIA Nadjet

PhD student at Sorbonne University
Team : NPA
https://perso.lip6.fr/Nadjet.Belblidia
https://perso.lip6.fr/Nadjet.Belblidia

Supervision : Serge FDIDA

Co-supervision : DIAS DE AMORIM Marcelo

Capacity Aware Opportunistic Networks: Characterization and Impact on the Dissemination of Large Contents

A common assumption in intermittently-connected (or opportunistic) mobile networks is that any contact has enough capacity to transfer any amount of data. Although such an assumption is reasonable for analytical purposes and when contents are small, it does not hold anymore when users produce contents that are larger than the contact capacity. Because wireless communications are generally omnidirectional, contact capacity relies on both contact duration and contact surrounding environment. Nevertheless, the literature on the design of communication protocols for opportunistic networks often assumes that contacts between nodes do not undergo the influence of other neighboring contacts.
In the first part of this thesis, we define and evaluate the surround indicator as a metric to exhibit a contact's surrounding environment in opportunistic networks. By using four real-world mobility datasets, we make three important observations: (i) contacts that present the same duration show significant differences in their surrounds, (ii) within a single contact the surround indicator can be extremely varying (indicating unstable surrounds), and (iii) avoiding highly surrounded zones can significantly enhance one-hop communication performance, in particular under high traffic load. Our proposal brings new insights into the topological characterization of opportunistic networks and we believe that the surround indicator can be a good approximation for potential interference whenever traffic information is not available.
Whereas communication opportunities are limited in terms of both duration and surrounding environment, users conversely generate, consume, and share contents that are becoming increasingly larger. In such a situation, content-sharing solutions must be reformulated to enable exchanging large contents. Users must slice data and send fragments separately, which leads to a better use of short-lived contacts and promotes progressive dissemination of large contents. The main question here is to design the best strategy for deciding which piece(s) to transmit whenever nodes meet. Furthermore, the piece size must be adequately selected before addressing this problem. Although small pieces imply a better use of short contacts, they generate more overhead due to the higher ration header/data in each piece. In the second part of this thesis, we investigate these two issues: piece size selection and piece selection strategy. First, we theoretically define the global goodput by pointing out the tradeoffs between turning small contacts into useful opportunities and decreasing piece overhead. Results from real-world traces show that, for reasonable header size, the piece size can be selected from a large range of values without significantly impacting the results. Second, we present the design and evaluation of PACS (Prevalence-Aware Content Spreading), a completely distributed algorithm that selects pieces to transfer based on a local view of their popularity. We evaluate the performance of PACS using both synthetic and real-world traces. When compared with sequential and randomized solutions, we show that PACS significantly outperforms these approaches both in terms of latency to achieve full dissemination and ratio of effective contacts. Moreover, PACS achieves performance levels that are extremely close to a centralized oracle version.
Finally, we present some experimental results obtained using PePiT, an Android application based on PACS that enables the dissemination of multimedia files among collocated Android devices in ad hoc mode. PePiT has been deployed on ten Android devices. Results show the practicability of PACS protocol on a real opportunistic network. Then, we go one step further in the investigation of large content dissemination challenges in opportunistic networks. We argue that uniform random inter-content selection may not be sufficient in real-world deployment. We propose EPICS, a distributed strategy that enables fulfilling dissemination policy objectives. EPICS is based on the grey relational analysis to weight content selection. In our study, we use EPICS to reduce the dissemination delay variability due to the uniform random inter-content selection. The goal is to ensure more homogeneous content dissemination delays for all contents regardless of their creation time and size. Based on the same testbed, we evaluate the improvement of EPICS. Thanks to a more appropriate content selection, pacsstar alleviates content dissemination delay disparities.

Defence : 04/12/2012

Jury members :

André-Luc Beylot, Professeur ENSEEIHT [Rapporteur]
Thierry Turletti, Directeur de Recherche INRIA [Rapporteur]
Michel Diaz, Directeur de Recherche CNRS
Serge Fdida, Professeur UPMC Sorbonne Universités
Franck Legendre, Chercheur ETH Zurich
Marcelo Dias de Amorim, Directeur de Recherche CNRS

Departure date : 09/01/2012

2009-2015 Publications