A Deep-Learning Enabled Traffic Analysis Engine for Video Source Identification
Yan Rong Shi, Subir Kumar Biswas · 2019
This paper presents a deep-learning based traffic analysis system for identifying streaming video sources within encrypted tunnels. The overall goal is to extract information about applications and traffic sources for system administrators and internet service providers to perform better resource allocation and other traffic management operations. The proposed architecture uses a deep neural network (DNN) for traffic analysis-based video source classification. The DNN model is trained and validated using an experimental setup that simulates real-world complex situations such as traffic combinations within encrypted VPN tunnels. The implemented DNN classifiers are shown to provide acceptable classification accuracies for single and combined multi-stream scenarios when trained with our proposed features.