Lightweight Traffic Classification and Forwarding Scheme for Software Defined Networks(SDN's)

Abstract

Ever since the boom of social media, the internet has drastically changed our lives. From social media platforms, Industry 4.0, Big Data, Cloud Computing to Augmented and Virtual reality, the amount of data on the internet has increased on an unprecedented scale, and it is not going to decrease in the years to come. This increasing amount of data puts a huge strain on today’s network infrastructure. Internet Network Traffic Classification is thus a new paradigm in which lightweight sophisticated algorithms are designed and tested on live networks. These algorithms help in setting the forwarding rules in a controller in a Software-Defined-Network based environment. In regard to increasing audio and video streaming services, this problem has been focused a lot with the help of Machine and Deep learning. These models do provide a high accuracy of classification, but can become complex problems depending upon the specific application. This paves the way for the need of practical, trivial architectures. Thus, in this thesis, a novel Fuzzy Logic based Lightweight Classification Architecture is proposed and implemented to successfully identify and segregate Audio and Video from other Network Traffic (NT) applications. First, a traffic classification scheme is devised and formulated using a basic handful of features. The details of these features from a variety of applications including audio and video is elaborated. These features are used later on to build the fuzzy based classification model. The classification model will then distinguish audio and video based interactive and streaming services from other random internet traffic based on rules set by Fuzzy that mimic human intuition. To verify the proposed algorithm, flow-based network captures is used to identify whether the correct class of network traffic is predicted or not. The results obtained from the Fuzzy logic is also compared with traditional Polynomial regression and kNN based Machine Learning algorithms to showcase the superiority of Fuzzy logic. Keywords: Internet Traffic Classification, Fuzzy Logic