网络切片是 5G网络的关键技术,在支持多种 5G应用和服务方面发挥着重要作用.为确保 5G网络提供更加灵活安全的按需服务,对网络切片的灵活性和安全性研究尤为重要.为此,提出一种按应用服务类型划分细粒度网络切片的方案,并通过基于卷积神经网络(Convolutional Neural Networks,CNN)的模型来安全分配网络切片.当网络流通过该模型后,先筛选出受到分布式拒绝服务(Distributed Denial of Service,DDoS)攻击的流量,然后良性流量再按应用类型分配到相应的切片上.仿真结果表明,基于CNN的网络切片分配模型,在安全分配网络切片方面有着显著的效果.与其它常见的机器学习分类算法相比,该方案中的模型在准确率、精确率、召回率和F1 分数方面都有着更好的性能优势.
Research on Secure Allocation of 5G Network Slices Based on CNN
Network slicing is a key technology for 5G networks and plays an important role in supporting multiple 5G applications and services.In order to ensure that 5G networks provide more flexible and secure on-demand serv-ices,the research on flexibility and security of network slicing is particularly important.Therefore,this paper proposes a fine-grained network slicing scheme divided by application service types,and securely allocate network slices through a model based on convolutional neural networks(CNN).After the network flow passes through the model,the traffic attacked by distributed denial of service(DDoS)is filtered out,and then the benign traffic is allocated to the corresponding slice according to the application type.The simulation results show that the network slice allocation model based on CNN has a significant effect on the secure allocation of network slices.Compared with other common machine learning classification algorithms,the model in this scheme has better performance advantages in terms of ac-curacy,precision,recall and F1 score.
Network slice securityConvolutional neural networkDistributed denial of service attack
NETL
NSTL
万方数据
