Basically, my research is oriented towards three main topics: Reflective Middleware, Mobile Systems and Critical Autonomous Systems.
During my thesis, I tried to overcome the lack of openness in middlewares such as Corba using open-compilers like open-c++ and open-java to provide transparency of faul-tolerance for the application programmer. I designed a meta-object protocol (MOP) well-suited to Corba applications. I used an open-compiler to apply this MOP at compile-time to regular Corba code. The compiled code then support the MOP and provides reification of its creation/destruction/methods invocation, intercession of the same events and introspection of its state. I developped on top of that an architecture that provides transparently several fault-tolerance mechanisms, like primary-backup replication, to standard Corba applications. One of the major conclusion of the thesis is that some information isn't available at the language level and is dependent on the lower-layers (operating system and/or middleware). This is why we decided to get involved into reflective middleware.
Francois Taiani's thesis was about applying the notion of reflection to complex systems whose architecture is based on several layers of software: operating system, middleware, etc. He defended and obtained his thesis in January 2004 and is currently a post-doc in New-Jersey ATT labs, in the USA. We are currently working on an implementation of these concepts .... more info here (Intranet only)
During my research fellowship at Trinity college in the Distributed Systems Group, We worked on new semantics of group communication for mobile participants. Basically, the idea is to incorporate the notion of proximity into the definition of groups. This led to the publication of two short papers.
At LAAS, we are currently working on the specific issues of fault-tolerance and more generally of dependability for mobile ad-hoc systems. We look at these problems from a middleware point-of-view as we don't see ourselves as network experts: what would be the usefull services to provide to help developpers building reliable ad-hoc systems. This is clearly a question which has not received satisfying answers yet. The MoSAIC (Mobile System Availability Integrity and Confidentiality) project officially started in September 2004. Its goal is to explore dependability issues of mobile applications at the middleware level. The project is partially financed through the French initiative for research in security and informatics (ACI S&I).
In 2003, Amine Filali CS Master work was to use the notion of obstacles (both motion and radio propagation obstacles) into wireless network simulations. He modified a network simulator called GloMoSim in order to take obstacles into account. Based on this work, he did some statistical experiments on three ad-hoc routing protocols (FSR, AODV and LAR) in three different obstacle scenarios : open-space, manhattan and an office. Here are his defense slides in french.
In 2002, Claudio Basile came at LAAS-CNRS for the summer. The work we did together is summarized in this survey : "A Survey of Dependability Issues in Mobile Wireless Networks" C. Basile, M-O. Killijian, D. Powell Technical Report, LAAS CNRS Toulouse, France, 2003.
We recently launched a new thesis on the subject of autonomous critical systems with the LAAS Robotics group. Benjamin Lussier has began in 12/03.