SDNCampus: Application Classification Dataset for SDN Enabled Campus Networks

Abstract

The exponential growth and sophistication of campus networks tailored with advanced applications has resulted into network congestion and Quality of Service (QoS) degradation. Meanwhile, the emerging paradigms of Software Defined Networking (SDN) and Deep Learning (DL) are promising technologies that enable network re-design and management in a centralized way. While SDN brings control of the entire network at the controller, DL can leverage statistical information offered by SDN to understand the characteristics of network applications better and inform the controller accordingly. However, deploying SDN coupled with DL in campus networks is still immature due to the lack of a proper dataset to train the DL models. This manuscript aims to generate an SDN specific dataset for application classification, including IoT application integrated in a campus network. Furthermore, the study demonstrates the usefulness of the proposed dataset by performing experimental evaluation using four (4) time series DL based techniques for traffic classification.