rubiruchi/machine-learning-in-traffic-classification-of-sdn-based-networks

Machine-learning-in-Traffic-classification-of-SDN-based-networks

1.Set up topology The nodes are set up in linear mode as shown below. Every node is basiclly an instance generated on cloud platform. They are assigned overlay addresses and configured as host and swiches by OpenFlow. Therefore they can be regarded as SDN. sw1 and sw2 are also connected to a controller

h1-sw1-sw2-h2(application installed)

h1 serves as a router connecting to the external network so that all the traffic generated by applications on h2 will be directed through h1.

2. Data collector The information of traffic flow are collected by controller connecting sw1 and sw2.

3. Training Supervised, unsupervised and semi-supervised learning algorithms are applied on dataset offline and models are exported.

4. Integration The integration of training models and controller make the real-time traffic classification possible.