Abstract:Encrypted traffic classification is of great significance for advanced network services. Though encryption methods seem unbroken in protecting users' privacy, existing studies have demonstrated that with sophisticated designed approaches utilizing the methods of machine learning or deep learning, the traffic can be identified as generated from which application type or even the specific application. However, most of the previous approaches either lack the generalization ability in different tasks or can hardly achieve the precise performance. One of the reasons is that they perform the classification from an incomplete perspective. To our best knowledge, none of which consider combing the payload content and payload statistics for encrypted traffic classification. Hence, in this paper, we propose the comprehensive effective traffic information analytics (CETAnalytics) framework to tackle the problem. Firstly, the comprehensive effective traffic information is specified and the motivation for combing the two aspects of the traffic is introduced. Based on the specification, the CETAnalytics framework utilizing the consolidated information and its devising implementation details are elaborated. Briefly, the implementation is totally built on the neural network for its high flexibility and powerful functionality to integrate the two dimensional analytics. Among the challenges tackled in the implementation, a substructure network Attract designed with the purpose of matching the traffic structure is proposed to realize the payload content analytics, which is one of the highlights of our work. For evaluation, several solid experiments are conducted using three designed tasks originated from the ISCXVPN2016 dataset. The experiment results show that (i) the effectiveness of the framework design for encrypted traffic classification; (ii) our implementation can achieve both high precision and robust generalization performance at the same time.

CETAnalytics: Comprehensive Effective Traffic Information Analytics for Encrypted Traffic Classification

CETP: A novel semi-supervised framework based on contrastive pre-training for imbalanced encrypted traffic classification

High-speed Encrypted Traffic Classification by Using Payload Features

An Encrypted Traffic Classification Framework Based on Higher-Interaction-Graph Neural Network

A Hypernetwork-based Personalized Federated Learning Framework for Encrypted Traffic Classification

Network Traffic Classification Model Based on Spatio-Temporal Feature Extraction

Seeing Traffic Paths: Encrypted Traffic Classification With Path Signature Features

Rethinking Encrypted Traffic Classification: A Multi-Attribute Associated Fingerprint Approach

EETC: an Extended Encrypted Traffic Classification Algorithm Based on Variant Resnet Network

Optimizing Feature Selection for Efficient Encrypted Traffic Classification: A Systematic Approach

Encrypted Traffic Classification Based on Traffic Reconstruction

IP Packet-Level Encrypted Traffic Classification Using Machine Learning with a Light Weight Feature Engineering Method

Multi-Task Scenario Encrypted Traffic Classification and Parameter Analysis

An encrypted traffic identification method based on multi-scale feature fusion

MATEC: A lightweight neural network for online encrypted traffic classification

End-to-end encrypted network traffic classification method based on deep learning

FastTraffic: A Lightweight Method for Encrypted Traffic Fast Classification

Encrypted TLS Traffic Classification on Cloud Platforms

An Novel Hybrid Method for Effectively Classifying Encrypted Traffic

A3C System: One-Stop Automated Encrypted Traffic Labeled Sample Collection, Construction and Correlation in Multi-Systems

Encrypted Traffic Classification Using Graph Convolutional Networks.