Description
Date depot: 1 janvier 1900
Titre: Probabilistic models for distributed systems and networks
Directeur de thèse:
Sébastien TIXEUIL (LIP6)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
A distributed system can be represented by a graph in which nodes
are the computing resources and edges are used for the communications
between nodes. In this context, a distributed algorithm is executed locally
by the nodes which use classical procedure to send and receive messages from
the other nodes. Examples of such systems are large-scale networks like
peer-to-peer networks, wireless sensor networks and mobile sensors (robots)
networks. For such networks, generally composed of a large number of nodes
and a complex structure, it is very difficult and even prohibitive to obtain
information on the network in a centralized manner. Moreover, the increase
in the number of nodes leads to a large dynamic behavior that may be caused
either by the nodes mobility (robot networks) or by the churn
(nodes may join or leave the system at will) for instance in peer-to-peer
networks.
These behaviors are clearly random phenomena that must be modeled in order
to evaluate the performance of the network.
For instance, in a sensor network, one must avoid the situation in which
some critical nodes, leave
the network. In the context of robot networks, the randomness comes from
the movement of the robots on a graph and it is thus important to evaluate
the time at which they meet each other. Moreover, it could be interesting to
insert some randomness into the distributed algorithm to improve the
performance of the system.
Doctorant.e: Maurer Alexandre