摘要
由于云网融合MCU多点控制过程中,受到时延影响导致视频会议系统信息流传输安全性较低。针对该问题,提出面向云网融合MCU联邦学习框架的安全性增强方法。设计虚拟网络功能映射联邦学习架构,最小化安全性增强时延。利用MCU的联邦学习架构,实现云计算环境下的协议辨识,结合4阶段流水的PF_RING处理模式分析网络流量使用情况。采用全同态加密方法,将明文加密处理后的数据存储在联邦学习框架中。通过拆分MCU联邦学习框架获取加密密文,以此增强MCU安全性。由实验结果可知,该方法漏洞个数与实际个数一致,流量使用情况与实际数值之间的误差为10 MB,具有高效的安全性增强效果。
Due to the impact of time delay in the multi point control process of cloud network integration MCU,the security of information flow transmission in video conferencing systems is relatively low.In order to solve this problem,a security enhancement method of MCU Federated learning framework for cloud network integration is proposed.Design virtual network function mapping Federated learning architecture to minimize security enhancement delay.Use the Federated learning architecture of MCU to realize protocol identification in the cloud computing environment,and combine the PF of 4-stage PF_RING processing mode analyzes network traffic usage.The full homomorphic encryption method is used to store the plaintext encrypted data in the federated learning framework.The security of MCU is enhanced by splitting the MCU Federated learning framework to obtain encrypted ciphertext.The experimental results show that the number of vulnerabilities in this method is consistent with the actual number,and the error between the traffic usage and the actual value is 10 MB,which has an efficient security enhancement effect.
作者
李楠
陈大卫
魏正荣
罗威
LI Nan;CHEN Dawei;WEI Zhengrong;LUO Wei(Information&Communication Company,State Grid Shanghai Electric Power Company,Shanghai 200040,China;Nanjing NARI Information&Communication Technology Co.,Ltd.,Nanjing 211100,China)
出处
《电子设计工程》
2024年第20期57-61,共5页
Electronic Design Engineering
基金
国网上海电力公司科技项目(52090F220001)。
关键词
云网融合
MCU
联邦学习
安全性增强
全同态加密
cloud network integration
MCU
federated learning
security enhancement
full homomorphic encryption
