基于联邦学习的无线通信网络DoS攻击检测方法

Method of wireless communication network DoS attack detection based on federation learning

无线通信网络受到DoS攻击,会使得网络的负载增加,导致延迟增加。而在无线通信网络中,数据通常分散在多个节点上,这会造成数据泄露和被攻击。为此,提出一种基于联邦学习的无线通信网络DoS攻击检测方法。对初始无线通信网络数据进行预处理和归一化,并采用随机森林算法进行降维处理,去除冗余特征,获得最佳网络数据特征集。将特征集输入到以深度卷积神经网络为通用模型的联邦学习训练模型中,独立训练本地模型并进行模型修正,传输至中心服务器进行聚合,收敛后完成训练。利用训练得到的联邦学习模型检测无线通信网络DoS攻击速率,再与接收者接收的容量最大值进行比较,判断是否有DoS攻击。实验结果表明,所提方法在处理大量数据时具有较高的稳定性和可靠性,能够在短时间内准确地检测出DoS攻击。

A DoS attack on wireless communication network can increase network load,resulting in more delay.In wireless communication network,data is usually scattered across multiple nodes,which can lead to data leakage and attack.Therefore,a method of wireless communication network DoS attack detection based on federation learning is proposed.The initial wireless communication network data is preprocessed and normalized,and the random forest algorithm is used for the dimensionality reduction to remove redundant feature and obtain the optimal network data feature set.The feature set is input into the federation learning training model with deep convolutional neural network as the general model,to independently train the local model and perform the model modification.It is transmitted to the central server for aggregation,and the training is completed after convergence.The trained federation learning model is used to detect the DoS attack rate in wireless communication network,and compared with the maximum capacity received by the receiver to determine whether there is a DoS attack.The experimental results show that the proposed method has high stability and reliability when processing large amounts of data,and can accurately detect DoS attack in a short time.