Software-Defined Networks separate the data plane from the control plane by transferring the intelligence from the forwarding devices to one or multiple controllers. The controllers can monitor the network and can deploy network-wide applications independently of the underlying equipment.
In parallel, the energy consumption of networking infrastructures is becoming a growing concern. For instance, networking devices represent 10 to 20% of the total energy cost of data centers. In most cases, proposed energy efficiency solutions suggest shutting down unused devices to reduce energy consumption in the network.
During my thesis, I leveraged SDN features to implement green policies while considering limitations such as forwarding table sizes of SDN switches. I also looked at the progressive deployment of SDN devices on existing legacy wired and optical networks. Furthermore, I studied the effects of Network Function Virtualization on energy savings. At the same time, on another theme, I also worked on structured distributed peer-to-peer systems for live video streaming.
During my postdoc at Concordia University, I continued my work on network virtualization, and I studied multi-layer network planning and resource and spectrum assignment (RSA) using classic optimization tools such as column generation.
As a research engineer at Huawei Technologies, I studied 5G network optimization problems. I worked on network slicing algorithms for hard isolation, multi-slice optimization, and I also leveraged machine learning tools and integrated them with linear programming methods.
Currently, as an associate professor at IMT Atlantique, I continue working on virtualized networks.