Description
Date depot: 13 mars 2019
Titre: Fundamental limits of machine-learning aided caching
Directeur de thèse:
Petros ELIA (Eurecom)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
The thesis will be about
large wireless communication networks, and its aim will be to develop the
theoretical and practical foundations of how we can apply preemptive use of
storage capacity at the nodes, to surgically alter the informational structure
of networks, making them faster, simpler and more efficient.
Phds work will seek to jointly view two
two approaches (feedback information theory, and coded-multiplexing) that were
thought to be disconnected; one uses feedback on the wireless PHY layer, the
other uses memory on the (mainly wired) MAC. He will explore evidence of a
powerful duality between the two, which can allow us to leverage memory, to
bypass the need for real-time-feedback, and to offer unprecedented throughput
gains. This partly motivates the thesis work of Adeel Malik which will seek to
provide a mathematical convergence of feedback information theory and
distributed storage.
In the end, I believe that in this thesis, if successful, the gains can indeed be notable, and they would come at
a time when current methods fail to address adequately the anticipated extreme
increase in users and demand.
Doctorant.e: Malik Adeel