Section: New Results

Pervasive computing

Participants : Laurent Ciarletta [ contact ] , Vincent Chevrier [MAIA Team], Tom Leclerc, Julien Siebert, Cyril Auburtin.

In Pervasive or Ubiquitous Computing, a growing number of communicating/computing devices are collaborating to provide users with enhanced and ubiquitous services in a seamless way. This is a domain that we are exploring in general and that can be considered at the convergence of several Computer Science fields such as Networking, Embedded Systems, Software engineering, and CHI/AI. Madynes is focussing on Ubiquitous networks and services, where there is a lot of different requirements and research topics that can be both generic or specific to our domain of expertise. But we are collaborating with other research teams (mainly INRIA - Maia) to be able to encompass issues and research questions pertaining to this domain in a wider way. The following lists those specifically related to the work done within Madynes:

• An adaptable yet high level of safety and security is needed. These computing devices should be working in such a way that common users can trust and rely on them,

• Pervasive Computing is high-technology seamlessly woven into our everyday life: therefore it requires autoconfiguration and reconfiguration of its elements and networks,

• The technologies need to be evaluated not only per domain, but on a larger scale, where end-user concerns and interactions should also taken into account. We are still pursuing our investigations on this domain and have been working more specifically on 3 prospects:

  • Multi-models of these Pervasive computing environments (including the users in the modelisation and the simulations). We have been focusing on the collaborative simulations of dynamic networks/elements, namely P2P (to be extended to adhoc networks) using agents to drive those simulations. This work is done in collaboration with the MAIA team.

  • State of the art on Service Discovery protocols, contextual metrics in adhoc networks, and Service Discovery in adhoc networks using an hybrid between cluster-like (WCPD) and MPR-based (OLSR) broadcasting [8] .

  • Geolocation of wireless devices: a research collaboration with Fireflies RTLS was started in March 2009.

Pervasive Computing is built around a user-centric model. In distributed, dynamic networks, services and applications, such as Peer-to-Peer (P2P) networks or Mobile Ad hoc NETworks (MANET), the users behaviour has a strong influence on the quality of service (QoS) and reciprocally.

We've first proposed to use models and simulators from both internetworking and AI (human behaviour) fields and then to make them interact rather than building one as an extension of the other.

Pervasive Computing is built around a user-centric model. In distributed, dynamic networks, services and applications, such as Peer-to-Peer (P2P) networks or Mobile Ad hoc NETworks (MANET), the users behaviour has a strong influence on the quality of service (QoS) and reciprocally.

We've first proposed to use models and simulators from both fields (internetworking and AI (human behaviour) and then to make them interact rather than building one as an extension of the other.

At first, an existing simulator (Peerfactsim) has been extended, with a strong coupling approach to study the influence of the rate of cooperation of user and the rate of pollution of data on the functioning of the P2P network. Results [11] showed the limits of such a strongly tied and centralized approach. A coupling of models and simulators has been proposed and we've first tackled the coordination issues (synchronization, compatibility and coherence) by using the Agent and Artefact paradigm. We have developped a decentralized coordination framework called AA4MM. The aim of this framework is to make heterogeneous simulators interact in such a way that coordination and integration issues are transparent for the people involved in the simulation process. When someone wants to include an existing simulator within the AA4MM framework, only few changes are needed. Moreover, the framework is based upon a decentralised coordination model that has been formalised (in Event-B) [45] in collaboration with Joris Rehm, in order to prove that coordination occurs with a finite number of simulators and that no deadlock is possible. Source code and JMS implementation have been developped in collaboration with Virginie Galtier Ciarletta from Supelec, Metz. Examples, demonstrations and the first realase of the framework are available(http://www.loria.fr/~siebertj/aa4mm/aa4mm.html ).his framework is currently used in oder to study the impact of the user mobility on the performances of MANET.

In order to build a stable yet adaptable architecture for context-aware service discovery with multimedia capabilities, we've looked into 2 ways of organising and communicating between mobile nodes. Previous work around the study of properties of clustering and adhoc protocols applied to contextual service discovery has been invited for an extended version in [8] where the number of cluster reorganisation has been improved in mix-mode mobility (similar proportion of highly moving nodes and more static ones). Following these results, a new protocol called SLSF (Stable Linked Structured Flooding) has been developed and is currently under submission for ICC 2010. It is based on clustering (WCPD) to allow for a quick toplogy building with an OLSR-like protocol for inter-cluster communication. It is enriched with a fault-recovery mechanisms. Overall, this solution gives a high-reachability with a low bandwidth usage. An architecture for advanced service discovery has been proposed [46] in the context of the ANR SARAH. In this project, the core of the hybrid network is a MANET, but a cloud of fixed servers and access-point surrounds this unstable core but can't be relied on. Our solution allows for an autonomous service discovery that is improved when the fixed services are available, and is currently being developed and tested wit our mulit-simulations and will hopefully be tested in real-scale at the Museum des Télécoms.

The collaboration with the University of Luxembourg has been extended to Collaborative Filtering in adhoc networks [17] , with P. Gratz. This is a technique to share and rate (multimedia) content in adhoc networks in a way that avoids flooding with information and requests, by filtering the "pertinence" of the exchanged data with regards to user profiles.