分布式网络数据异常结构优化识别仿真
|
摘要
对分布式网络数据异常结构的优化识别,能够有效解决计算机网络安全问题。对数据的异常数据进行识别,需要结合间隙统计方法,度量参考数据集与观测数据集之间的间隙聚类,完成数据异常结构的优化识别。传统方法得到数据流量中的流量关键点,将数据转换映射到相应关键点后建立直方图,但忽略了对数据集间隙聚类的度量,导致识别精度偏低。提出基于余弦聚类的识别方法。计算每个数据流特征的特征熵和特征比,选取异常结构识别的特征属性向量,根据余弦聚类的方法计算每个数据流特征样本到每个聚类中心的余弦距离,直到获得的聚类中心保持不变或变化较小;最后利用间隙统计方法,度量参考数据集与观测数据集之间的间隙聚类,实现异常结构的识别。实验结果证明,所提方法能够实现分布式网络数据异常结构的准确识别,且能够较好地满足实时识别的需要。
This article proposes a recognition method based on cosine clustering. The characteristic entropy and characteristic ratio of each data flow are calculated,and then the feature attribute vector of abnormal structure recognition is selected. After that the cosine distance from feature sample of each data flow to each clustering center is calculated based on cosine clustering until clustering center remained constant. Finally,the gap statistical method is used to measure the gap clustering between reference data set and observation data set. Thus,the abnormal structure is identified. Simulation results verify that the proposed method can realize the accurate recognition of anomaly structure of distributed network data,which can satisfy the real-time recognition.
This article proposes a recognition method based on cosine clustering. The characteristic entropy and characteristic ratio of each data flow are calculated,and then the feature attribute vector of abnormal structure recognition is selected. After that the cosine distance from feature sample of each data flow to each clustering center is calculated based on cosine clustering until clustering center remained constant. Finally,the gap statistical method is used to measure the gap clustering between reference data set and observation data set. Thus,the abnormal structure is identified. Simulation results verify that the proposed method can realize the accurate recognition of anomaly structure of distributed network data,which can satisfy the real-time recognition.
引文
[1]牛咏梅.基于分形理论的光纤网络流量异常检测技术[J].激光杂志,2016,37(5):89-92.
[2]王铭鑫,等.一种SDN中基于熵值计算的异常流量检测方法[J].电信科学,2015,31(9):83-89.
[3]危美林,等.面向异常流量检测的自适应抽样算法研究[J].计算机应用研究,2015,32(10):3052-3055.
[4]马林进,等.基于词袋模型聚类的异常流量识别方法[J].计算机工程,2017,43(5):204-209.
[5]李兴国,王俊峰.无线网络数据流量异常优化检测仿真研究[J].计算机仿真,2017,34(9):386-389.
[6]罗玲,殷保群,曹杰.基于sketch数据结构与正则性分布的骨干网流量异常分析与识别[J].系统科学与数学,2015,35(1):1-8.
[7]费金龙,等.基于云模型的网络异常流量检测[J].计算机工程,2017,43(1):178-182.
[8]邱卫,杨英杰.基于尖点突变模型的联动网络流量异常检测方法[J].计算机科学,2016,43(3):163-166.
[9]曹杰,殷保群,钱叶魁.基于流特性的网络流量异常检测研究[J].系统科学与数学,2015,35(10):1127-1134.
[10]陈长辉.对等网络流量信息结构异常的检测技术研究[J].应用激光,2017,40(1):134-138.
[2]王铭鑫,等.一种SDN中基于熵值计算的异常流量检测方法[J].电信科学,2015,31(9):83-89.
[3]危美林,等.面向异常流量检测的自适应抽样算法研究[J].计算机应用研究,2015,32(10):3052-3055.
[4]马林进,等.基于词袋模型聚类的异常流量识别方法[J].计算机工程,2017,43(5):204-209.
[5]李兴国,王俊峰.无线网络数据流量异常优化检测仿真研究[J].计算机仿真,2017,34(9):386-389.
[6]罗玲,殷保群,曹杰.基于sketch数据结构与正则性分布的骨干网流量异常分析与识别[J].系统科学与数学,2015,35(1):1-8.
[7]费金龙,等.基于云模型的网络异常流量检测[J].计算机工程,2017,43(1):178-182.
[8]邱卫,杨英杰.基于尖点突变模型的联动网络流量异常检测方法[J].计算机科学,2016,43(3):163-166.
[9]曹杰,殷保群,钱叶魁.基于流特性的网络流量异常检测研究[J].系统科学与数学,2015,35(10):1127-1134.
[10]陈长辉.对等网络流量信息结构异常的检测技术研究[J].应用激光,2017,40(1):134-138.