Unknown network traffic classification method based on metadata assistance and federated learning

Abstract

The invention discloses an unknown network traffic classification method based on metadata assistance and federated learning, and the method comprises the steps: obtaining real unknown network traffic, and extracting real metadata in the real unknown network traffic; wherein the real metadata is partial byte data of the network flow data; inputting the real metadata into a federated learning global model, and outputting the category of unknown network traffic through the federated learning global model; wherein the federated learning global model is obtained based on a plurality of trained binary classifiers. According to the embodiment of the invention, the unknown network traffic is classified and identified according to the federal learning global model based on the metadata and the binary classifier, so that a client with the unknown network traffic can learn the classification method of the unknown traffic from other clients under the condition of protecting data privacy and security. 本发明公开了一种基于元数据辅助和联邦学习的未知网络流量分类方法，所述方法包括：获取真实未知网络流量，并提取所述真实未知网络流量中的真实元数据；其中，所述真实元数据为网络流量数据的部分字节数据；将所述真实元数据输入至联邦学习全局模型，通过所述联邦学习全局模型输出未知网络流量的类别；其中，所述联邦学习全局模型基于若干已训练的二分类器得到。本发明实施例基于元数据和二分类器，根据联邦学习全局模型来对未知网络流量进行分类识别，使得本发明能够一个具有未知网络流量的客户端在保护数据隐私和安全的情况下能从其他客户端学习到未知流量的分类方法。