基于BP神经网络的DDoS攻击自主检测方法

An Autonomous Detection Method for DDoS Attacks Based on BP Neural Network

分布式拒绝服务(Distributed Denial of Service,DDoS)攻击在网络中较为常见,但普通的DDos攻击检测方法难以对其追踪和防范,无法充分地考虑算法误差调整参数,导致检测精度较低。为此,提出基于反向传播(Back Propagation,BP)神经网络的DDos攻击自主检测方法,分析DDos攻击特点,采用信源地址、目标地址、包协议等数据包信息,提取DDoS攻击网络特征。采用误差BP算法进行参数训练,采用梯度下降法对各参数进行更新,利用BP神经网络进行DDos攻击自主检测。实验结果表明,通过对DDoS攻击的检测,该方法的检测准确率达到93.87%,并且具有良好的泛化性能。

Distributed Denial of Service(DDoS)attacks are common in the network,ordinary DDoS attack detection methods are difficult to track and prevent them,and cannot fully consider algorithm error adjustment parameters,resulting in low detection accuracy.Therefore,an autonomous detection method of DDoS attacks based on Back Propagation(BP)neural network is proposed,and the characteristics of DDoS attacks are analyzed.Data packet information such as source address,target address and packet protocol are used to extract the characteristics of DDoS attack network.The error BP algorithm is used for parameter training,the gradient descent method is used to update the parameters,and the BP neural network is used for DDoS attack autonomous detection.The experimental results show that the detection accuracy of this method reaches 93.87%through the detection of DDoS attacks,and has good generalization performance.