QR-SDN: Reinforcement Learning for Direct Flow Routing in Software-Defined Networks
Folder description
-
/controller
: Includes the controller implementation. Scripts can be started with:ryu-manager remote_controller.py
controller/config.py
: Config file containing all the parameters for the learning agent/controller (e.g. exploration strategy)
-
/mininet
: Includes the Mininet emulation scripts with different topologies. Scripts can be started with:sudo python Four_switches_two_ways.py
-
/plots
: Includes the measurement data and plot scripts. Scripts can be started with:python3 plot_<scenario>.py
-
/vagrant
: Includes Vagrantfile to spawn multiple VMs automatically and scripts to start and download measurements. Scripts can be started with:vagrant up rl_routing
,vagrant ssh rl_routingX
TODO's
- add Jupyter notebook to produce plots
- improve documentation