云边环境下微服务通信网络入侵行为识别方法  

Intrusion Behavior Identification Method of Microservice Communication Network in Cloud-Side Environment

作　　者：任华新[1] REN Huaxin(Liaoning University of International Business and Economics,Dalian 116033,China)

机构地区：[1]辽宁对外经贸学院,辽宁大连116033

出　　处：《通信电源技术》2024年第9期139-141,共3页Telecom Power Technology

基　　金：辽宁省教育厅科研项目(W2020JC005);辽宁省普通高等教育本科教学改革研究项目(2022SJJGZY12)。

摘　　要：通信网络入侵行为具有多样性,难以保障入侵行为的识别效果,因此提出一种云边环境下微服务通信网络入侵行为识别方法。引入非线性变换函数,对微服务通信网络入侵行为的特征进行半监督学习,通过将原始入侵行为映射到高维空间,获取正常运行状态下特征向量的参数。在构建以改进降噪自编码网络为基础的网络入侵行为识别模型的过程中,引入注意力机制,根据网络状态数据的重构误差与所设定的网络入侵行为特征阈值之间的差异,判断具体的入侵行为。根据测试结果,该方法能够准确识别出不同类型的入侵攻击。The intrusion behavior of communication network is diverse,and the effect of intrusion behavior identification is difficult to guarantee.Therefore,a method of intrusion behavior identification of micro-service communication network in cloud environment is proposed.The nonlinear transformation function is introduced to semisupervised learn the intrusion behavior characteristics of micro-service communication network,and the parameters of feature vectors are obtained by mapping the original intrusion behavior to high-dimensional space.The attention mechanism is introduced into the network intrusion behavior identification model based on the improved noise-reducing self-coding network,and the specific intrusion behavior is judged according to the difference between the reconstruction error of network state data and the set threshold of network intrusion behavior characteristics.In the test results,different types of intrusion attacks can be accurately identified without false positives.

关 键 词：微服务通信网络 入侵行为识别 非线性变换函数 半监督学习 注意力机制 

分 类 号：TN918[电子电信—通信与信息系统]