Multimedia transmission rules and encrypted audio and video traffic identification algorithm implemented in P4
Date
2021
Authors
Lu, Jiping, author
Partridge, Craig, advisor
Gersch, Joseph, committee member
Hayne, Stephen, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
With Internet traffic exponentially growing, it is critical for operators to identify voice call and video conference traffic and ensure their quality. Yet, the widespread deployment of encryption protocols makes it challenging to classify encrypted traffic. This research achieves a significant discovery of audio and video traffic transmission rules. Based on the rules, this paper proposes a general audio and video traffic identification algorithm. In order to evaluate this algorithm's performance, we designed and implemented an audio and video traffic identification algorithm in P4 (Programming Protocol-Independent Packet Processors). This promising research reveals that this algorithm achieves 98.98% accuracy on voice call identification. For video conferences, this algorithm achieves 96.24% accuracy on audio data identification and 88.75% accuracy on video data identification. Compared to current pervasive machine learning-based traffic classification approaches, this innovative algorithm bypasses complicated machine learning processes by directly applying audio and video traffic transmission rules on network functions, consuming less computation and memory resources.
Description
Rights Access
Subject
encrypted voice and video traffic identification
audio and video traffic transmission rules
multimedia traffic identification and classification