DIONYSOS  : Dependability, Interoperability and PerfOrmaNce analYsiS of netwOrkS

The main objectives of the project are the identification, the design and the selection of network architectures for communication services, as well as the development of computing methods and mathematical tools for the fulfillment of these tasks. These objectives lead to two types of complementary research fields: the systems' qualitative aspects (e.g. protocol testing) and the quantitative aspects which are essential to the correct dimensioning of these architectures and the associated services (analysis of performance, of dependability, of Quality of Service, of Quality of Experience (QoE), and of performability).

• The team works on the problems related to the study of complex communication systems using analytical techniques, numerical methods and Monte Carlo approaches, in performance and dependability evaluation. Regarding simulation, its activities have as main objective the development of new estimation techniques and the analysis of their properties. The team is also active in the Quasi-Monte Carlo area. Concerning analytical approaches for the study of Markovian systems in particular, we work on the evaluation of (very) large models, on the transient analysis of the associated stochastic processes, and on the analysis of some continuous (fluid) of hybrid (with both continuous and discrete parts) models.
• DIONYSOS has a specific research activity on the “direct” evaluation (as opposed to the analysis through models) of communication systems, from both qualitative and quantitative viewpoints. In the first case, our work concerns interoperability tests of components (hardware and software) around the IPv6 protocol. In the second case, the goal is the automatic evaluation of the perceptual quality of multimedia communications, in both one or two directions, audio or video, if necessary in real time, on the Internet. The perceived quality is the main component of Quality of Experience (QoE) in these strategic contexts.
• From the networking viewpoint, we work on the design of optical architectures (access and core networks), including fault tolerant aspects, and on wireless systems, more precisely on mobility management (oriented towards reliability) and on energy management in sensor networks. We also work on the use of Machine Learning techniques in solving different problems related to measuring or to resource management, always on communication networks. Our activities in QoE are also applied to the design of P2P architectures, and more specifically of Content Delivery Networks. Last, we have research activities on network economy, with the use, among other techniques, of game theory.