Project SARAH covers four main research topics:
Our work aims at supporting communication in disconnected mobile ad hoc networks (or MANETs for short). To achieve this goal we develop a flexible Java-based middleware platform called DoDWAN (for Document Dissemination in Wireless Ad hoc Networks), which leverages on the combined principles of opportunistic networking and delay-tolerant networking in order to account for the absence of end-to-end connectivity in disconnected MANETs.
DoDWAN was designed so as to be flexible, and easily extensible. It can notably accommodate different kinds of communication protocols. To date our effort has mostly been focused on supporting content-based communication in disconnected MANETs. DoDWAN therefore implements a protocol we designed along this line.
This work package is concerned by the service-oriented approach applied to disconnected ad hoc networks. Our aim is to provide the application programmer with a service platform that take into account all the specificities of disconnected MANETs. This service platform is intended to cover the service discovery aspects but also the invocation aspects. The two main characteristics that have to be taken into account are the communication delays and the absence of end-to-end connectivity, that often lead to an unpredictable availability of the service providers.
We have designed a first prototype of this service platform based on the DoDWAN communication support, using its publish/subscribe API. Preliminary experiments with this platform show that simple middleware mechanisms can be implemented at the service-level in order to cope with the induced delays on both the provider and the client sides. Clients can be provided with means to be aware of the available services network-wide (discovery) and they can asynchronously interact with these discovered providers (invocation).
Let us consider an open, delay-tolerant mobile ad hoc network, where mobile devices can use services and exchange information thanks to the benevolent help of —possibly initially unknown— neighboring devices. A malvolent device could easily claim a false or stolen identity or attempt to corrupt the integrity of the messages it receives before possibly forwarding these messages to the rest of the network.
The goal of this workpackage is to define new flexible and extensible security mechanisms that address these major issues. Special attention will be paid to the two following facts: (1) no permanently accessible service can ever be assumed and thus the standard solutions based on a central authority are inoperant; (2) the asynchronism inherent to delaytolerant networking imposes new constraints on the design of security related protocols.
Up to now the goal of most research works on the simulation of mobile ad hoc networks has been to closely model the behavior of the lowest layers of the network stack, and especially the MAC layer. In contrast the simulation of other aspects of mobile networking is often neglected. For example, most simulators implement very crude mobility models and little attention has been paid so far to the simulation and to the evaluation of the distributed algorithms that are meant to reside on top of the network stack.
The goal of this workpackage is thus to devise models and a simulation suite that will make it possible to simulate an environment composed of heterogeneously distributed mobile devices, as well as delaytolerant communication between these devices, delaytolerant service discovery and high level algorithms running on top of these devices. This package is based on the integration of the Madhoc and the DA-GRS simulators previously designed and implemented by the partners of the project.