Troubleshooting for Network Operators

3Traffic Classification: Novel QUIC Traffic Classifier Based on Convolutional Neural Network

“Predicting the future isn’t magic, it’s artificial intelligence”

Dave Waters

In Chapter 3, we present a traffic classification module to identify the application class for QUIC traffic. The application class also plays an important role in application-aware remediation approaches in network troubleshooting. This module is briefly described in the novel troubleshooting framework (Figure 2.2).

3.1. Introduction

Traffic classification (Lopez-Martin et al. 2017) plays an essential role in network troubleshooting for NOs. The objective of traffic classification is to identify application classes in the network to implement application-aware mechanisms to address network problems and meet strict SLA requirements. Concretely, traffic classification is used widely in application-aware routing, billing policies, intrusion detection systems, identifying security threats, enforcing QoS requirements, etc. (Rezaei et al. 2019). Therefore, traffic classification is studied thoroughly by the research community.

In the past, many attempts to identify the application class relied on port-based approaches and payload-based approaches. The first approach uses port numbers in TCP/UDP packets to identify the application class. However, it is not effective because modern applications and protocols are not always tied to a specific port. Besides, many networks change the port number in the packets using ...