Software, Prototypes and Demonstrators
The CDA team coordinates or participates to several research projects.
The CDA team coordinates ANR CIS 07 project CIP (ANR-07-CIS7-011). The CIP project aims at providing a set of tools for for high performance peer to peer computing, whereby each node that participates to the application is client and server of the application. We concentrate particularly on applications related to numerical simulation and optimization that present task parallelism and need solution via distributed iterative methods.
The CDA team is responsible of the design of the P2PDC decentralized environment for high performance peer to peer computing. The enviroment P2PDC facilitates the implementation of distributed algorithms and permits one to have direct communication between peers. The P2PDC decentralized environment facilitates programming thanks to a reduced set of communication operations and the automatic and self adaptive management of communication mode between peers via a self adaptive communication protocol. The self adaptive communication protocol based on the Cactus framework manages communication mode between peers in function of context elements of the network level like adherence to a cluster. Experiments have ben carried out on Grid 5000, the NICTA and PlanetLab testbeds. (see Publications and CIP). Partnairs of the CIP project are: LIFC, IRIT-ENSEEIHT, MIS and EMD. We collaborate also with NICTA, Sidney Australia (see CIP).
The CDA team participates to ANR PSIROB 06 project Smart Surface (ANR-06-ROBO-0009-03). The goal of this project is to design a distributed smart surface for sorting, conveying and positionning micro parts. The smart surface is based on MEMS technology. Each micro module will integrate sensors, processing unit and actuators. The cooperation of micro modules permits the smart surface to accomplish the differents tasks. The CDA team works on the design of distributed algorithms for global state acquisition of the smart surface, i.e. parts that lie on the smart surface. A specificity of the work relies on the low definition of the parts since the parts are tiny and few sensors are available. Several distributed synchronous and asynchronous algorithms have been proposed, their convergence has been established, stopping criteria have been given. Concurrent pattern recognition algorithms have been given. A smart surface simulator has been developed (see reference).
Team CDA is partner of ANR project Smart Blocks (ANR 2011 BS03 005) that started September 1, 2011. Project Smart Blocs aims at moving and positionning fragile tiny parts via a MEMS based distributed system that can reconfigure itself dynamically.