
I received my PhD degree in Computer Science from the university of Lorraine in 2010. Then I spent one year and a half in the ASAP team at the university of Rennes. Currently, I am a postdoctoral researcher in the ND group of the university of Oslo.
My research interests lie in the field of Distributed Systems. More specifically, I'm interested by very large scale systems such as Peer-to-Peer network. In these systems, I focus mainly on the problems of data consistency for collaborative editing, message propagation, publish/subscribe systems and, online social networks.
Collaborative editing allows users distributed in time and space to edit the same documents. Nowadays, collaborative editing becomes massive: a large number of participants can quickly have an important content in terms of quantity but also quality. For scalability concerns, I'm interested in Peer-to-Peer collaborative editing.
The basic problem is to provide algorithms to maintain data consistency compatible with the constraints of collaborative editing as well as Peer-to-Peer network constraints.
In this context, I have proposed Logoot, an optimistic replication algorithm dedicated to text document. Logoot is highly scalable, and can handle millions of users. The time complexity is O(log(n)), with n the number of elements in the documents. Logoot has been validated through experiments using modifications performed on Wikipedia.
Many distributed systems rely on communication channel with several level of properties ensured. For instance, most of Collaborative Editing Systems assume the existence of a reliable causal broadcast. Such a mechanism ensures, in short, that all peers receive all messages in an order following the "happened-before" relationship. Prior works on causal broadcast assumes that the system is a "closed-group", i.e., at one time, each peer knows the complete list of peers in the system. Other works use a sequencer to enforce the causal order. In both case, these approaches are not suitable for Peer-to-Peer network.
To provide a causal broadcast for Peer-to-Peer systems, I am currently working on a scalable causal broadcast.
Content-based publish/subscribe systems are a promising prospect for handling massive event streams. This communication paradigm offers rich filtering and allows to decouple the producers of data from their consummers.
I am currently interested in systems exploiting the social links between users in order to provide augmented services.
My publications can also be found on hal and google scholar.
I have provided 490 hours of teaching divided courses of