Networks and Distributed Software Systems

The Distributed Software Systems and Networks (RSLR) group is concerned with the "proper" functioning of distributed computer systems, which are at the heart of today's digital ecosystem and the societal changes we are experiencing.

In line with LISTIC's core theme, distributed computer systems are, on the one hand, the execution media for numerous information fusion systems. On the other hand, the very operation of these systems can benefit from the use of information fusion methods.

Our main focus is on the architectural, conceptual, operational and software aspects of these systems in all their application scenarios.

Our current and recent activities, supported by various national and international projects, focus on three complementary dimensions.

The first area focuses on IT systems and related issues, in terms of design, development, deployment, control and operation.

A second axis extends the limits of the system to the hardware, the execution medium and the interaction between the computer system and its execution medium.

A third axis takes into account the human user as a stakeholder in the system, both in the extraction of his needs and system specifications, and in the dynamic interaction(i.e. whose properties evolve during operation) between the system and its user; this interaction produces behavioral changes in the user and changes in the use and purpose of the system.

More specifically, we are interested in :

• and analysis of software-intensive, adaptable and dynamically evolving systems. Our work proposes approaches, languages and platforms for describing, deploying, executing, dynamically evolving and monitoring applications/systems based on services and/or components (SOA, Cloud, information systems, etc.). These approaches also coverbusiness process engineering. The languages and models proposed are based on the notion of orchestration (of services, software components/entities, activities/tasks for processes). These models are also at the heart of one of our major activities, which focuses on the analysis and runtime control of applications deployed in the Cloud;
• adaptation between processing and computing architecture. This adaptation is based on application problems and hardware platforms at various levels of granularity. For example, we are interested in high-performance processing chains for large-scale satellite image processing or on-the-fly video processing, whose execution support ranges from GPU components, through multi-core architectures, to clusters and grids, and whose granularity level is mainly the execution of a high-level task. But also to applications with finer levels of granularity, such as data flow or packet processing in networks, where hardware parallelism of GPU or many-core, FPGA or TCAM type must be exploited at algorithmic level to adapt processing to different heterogeneous hardware architectures.
• understanding the interaction between users and IT systems, but also the interaction between users through the medium of IT systems. This interaction generally takes the form of a dynamic loop, where a computer system is initially designed to meet certain user needs, but as users (whether end-users or intermediaries building new services on top of existing ones) appropriate the system, they develop innovative uses that were not initially foreseen. As these new uses and new user needs result in changes to the system, we observe a two-way interaction in which the user defines the framework of the IT system, but the IT system also changes the user's behavior. These interactions are fundamental to understanding the development of major IT systems such as social networks, crowd-sourcing and even the Internet.