基于人工蜂群算法的计算机网络DDoS攻击检测方法

Computer Network DDoS Attack Detection Method Based on Artificial Bee Colony Algorithm

计算机网络在DDoS入侵下容易出现停止服务、网络崩溃,为了提高网络安全性,提出基于人工蜂群算法的计算机网络DDoS攻击检测方法;基于特征样本之间的相关性,获取DDos攻击的同态样本分布时间序列输出,并且得到对网络入侵数据集的输出阵列模型,构建计算机网络DDoS攻击的自适应的入侵检测信息分析模型,根据网络数据流与潜在空间之间的映射关系,结合测试样本和学习样本之间特征差异性进行DDoS攻击数据特征提取,在基站上设置入侵检测数据处理终端,采用人工蜂群算法实现对计算机网络攻击检测的个体最优值和全局最优值寻优,获取在相对维度较高的网络入侵检测数据集中的入侵特征分布模型,计算特征选择的信息增益,得到快速相关性过滤输出,通过蜂群算法,分别选择不同的特征子集得到DOS攻击检测的离散信息,实现对组合网络流量数据间的攻击信息特征提取和聚类分析,解决计算机网络DDoS攻击检测过程中的连续多变量优化问题;仿真测试结果表明,采用该方法进行计算机网络DDoS攻击检测的寻优能力较好,精度和效率高于传统方法。

The computer network is prone to stop service and network crash under the intrusion of DDoS.In order to improve the network security,a DDoS attack detection method of computer network based on artificial bee colony algorithm is proposed.Based on the correlation between feature samples,the output of homomorphic sample distribution time series of DDos attacks is obtained,and the output array model of network intrusion data set is obtained.Then,the adaptive intrusion detection information analysis model of computer network DDoS attacks is constructed.The feature extraction of DDoS attack data is carried out based on the feature differences between test samples and learning samples.The intrusion detection data processing terminal is set up on the base station,and the artificial swarm algorithm is used to optimize the individual and global optimal values of computer network attack detection,so as to obtain the intrusion feature distribution model in the network intrusion detection data set with higher relative dimensions.The information gain of feature selection is calculated,and the output of fast correlation filtering is obtained.The discrete information of DOS attack detection is obtained by selecting different feature subsets through the hive algorithm,which realizes the feature extraction and cluster analysis of attack information among the combined network traffic data,and solves the continuous multivariate optimization problem in the process of DDoS attack detection on computer networks.The simulation test results show that this method has better optimization ability,higher accuracy and efficiency than the traditional method.