摘要
由于可见光通信网络恶意代码的升级和变形,导致恶意代码识别正确率与效率下降,因此,提出一种基于深度森林的可见光通信网络恶意代码识别方法。将恶意代码二进制文件映射为图片形式,并且通过分析图像梯度获取方向梯度直方图的特征,提取恶意代码图像特征向量。在随机森林模型中,利用决策树划分恶意代码类别,同时多粒度扫描图像特征向量矩阵,扫描结果作为深度森林的输入,输出中平均值最大的类别即为恶意代码,实现识别恶意代码。仿真实验结果表明,所提方法的识别误报率最低、正确率最高,并且识别效率最高,提高了光通信网络的安全性,其具有可行性和有效性。
Due to the upgrading and deformation of malicious code in visible light communication network, resulting in a decline in the accuracy and efficiency of malicious code identification, Therefore, a malicious code identification method based on deep forest in visible light communication network is proposed. The binary file of malicious code is mapped into the form of picture, the directional gradient histogram feature is obtained by analyzing the image gradient, and the feature vector of malicious code image is extracted. In the random forest model, the decision tree is used to divide the categories of malicious code, and then the image feature vector matrix is scanned at multiple granularity. The scanning results are used as the input content of the deep forest. The category with the largest average value in the output results is malicious code, which realizes the identification of malicious code. Simulation results show that the proposed method has the lowest false positive rate, the highest accuracy, and the recognition efficiency is the highest, improve the security of optical communication network, its feasibility and effectiveness.
作者
杨静
郭韦昱
杨文彬
YANG Jing;GUO Weiyu;YANG Wenbin(Taiyuan Normal University Network&Information Center,Jinzhong 030619,China;Central University of Finance and Economics Information school,Beijing 102206,China;Taiyuan Normal University Computer Department,Jinzhong 030619,China)
出处
《激光杂志》
CAS
北大核心
2022年第10期150-154,共5页
Laser Journal
基金
国家自然科学基金青年项目(No.6210022337)。
关键词
深度森林
可见光通信网络
恶意代码
决策树
多粒度扫描
deep forest
visible light communication network
malicious code
decision tree
multi granularity scanning
