Developing a Fake Classifier Node for DIFFUSE

DIFFUSE (Distributed Firewall and Flowshaper Using Statistical Evidence) is a network prioritisation system which implements Machine-Learning (ML) based techniques to classify and subsequently prioritise flows. Inefficient testing methodologies impacts on future development, making it difficult to upgrade and expand. We have developed a Fake Classifier Node (FCN) that enables manual creation of DIFFUSE messages, bypassing the ML components, leading to more efficient testing and more reliable results. Tests confirm that the FCN is able create DIFFUSE messages and that they are correctly processed by existing DIFFUSE Action Nodes.