A distributed traffic classification model based on Federated Learning

Abstract

Traffic classification is the basis of protocol and application design for high performance networks, and the guarantee of network operation management, network traffic scheduling and network development planning. There are some problems in current traffic classification methods, such as difficulty in obtaining and labeling actual traffic data, legal risk and so on. In order to solve these problems, we propose a traffic classification model generation method based on federated learning.This work verifies the method to get the traffic classification model by collecting and processing the traffic at the user terminal and training it based on Federated learning under the premise of protecting the privacy security of users. Experiments show that the accuracy of the traffic classification model based on federated learning reaches 96.4%.