基于ESN神经网络的光通信网络安全态势辨识研究

Research on security situation identification of optical communication network based on ESN neural network

光通信网络在传输信息时,很容易被非法攻击,发生信息窃取、篡改和删除等泄露行为。针对上述问题,为保证光通信网络安全,提出一种基于ESN神经网络的光通信网络安全态势辨识方法。该方法采用NetFlow技术设计采集器,采集NetFlow流量数据,并实施离散化处理。将NetFlow流量数据转换为流量灰度图像,并借助灰度共生矩阵提取图像特征,包括像素灰度分布的均匀程度、图像包含的信息量、图像的视觉清晰度、灰度共生矩阵元素排列的相似程度、图像局部灰度均匀性,作为NetFlow流量数据的特征。以5项特征为输入,利用ESN神经网络构建辨识模型,得出光通信网络安全态势类型。结果表明:与基于卷积神经网络的识别方法、基于贝叶斯的识别方法以及基于随机配置网络的识别方法相比,所研究方法应用下的杰卡德系数更高,说明该方法辨识准确性更高。

When transmitting information,optical communication network is easy to be illegally attacked,and in-formation theft,tampering and deletion occur.Aiming at the above problems,in order to ensure the security of optical communication network,a security situation identification method of optical communication network based on ESN neu-ral network is proposed.This method uses NetFlow Technology to design the collector,collect NetFlow flow flow data,and implement discrete processing.The NetFlow traffic data is converted into traffic gray image,and the image features are extracted with the help of gray level co-occurrence matrix,including the uniformity of pixel gray distribution,the amount of information contained in the image,the visual clarity of the image,the similarity of gray level co-occurrence matrix element arrangement,and the local gray level uniformity of the image,which are the features of NetFlow traffic data.Taking five features as inputs,the identification model is constructed by ESN neural network,and the security situation type of optical communication network is obtained.The results show that compared with the recognition meth-od based on convolution neural network,Bayesian recognition method and random configuration network,the jackard coefficient of the research method is higher,which shows that the recognition accuracy of this method is higher.