基于IPv6技术的局域网入侵检测系统设计  

Design of LAN intrusion detection system based on IPv6 technology

作　　者：胡巧婕 刘永辉 HU Qiao-jie;LIU Yong-hui(Student Management Office,Xinjiang Changji TV and Radio University,Changji 831100,Xinjiang Uygur Autonomous Region,China;Information Management Office,Xinjiang Changji TV and Radio University,Changji 831100,Xinjiang Uygur Autonomous Region,China)

机构地区：[1]新疆昌吉广播电视大学学生管理处,新疆昌吉831100 [2]新疆昌吉广播电视大学信息管理处,新疆昌吉831100

出　　处：《信息技术》2022年第9期46-50,共5页Information Technology

基　　金：新疆维吾尔自治区教育厅高等职业院校创新和发展教学改革项目(XJ16-006)。

摘　　要：针对局域网IPv6协议数据入侵检测的异常问题,文中设计并实现了基于神经网络的IPv6局域网入侵检测方法。利用深度学习中的前馈神经网络建立模型,将入侵行为所导致的异常数据流作为原始输入驱动训练模型,并结合误差逆传播与优化函数对系统做性能优化。在校园网内进行的实验结果表明,与经典的基于模式匹配算法的入侵检测方法相比,文中所提方法在正确率和误报率方面均优于基于模式匹配的算法,且在相同条件下,比基于模式匹配的算法在检测数量方面至少超出5%。Regarding the abnormal problem of data intrusion detection of IPv6 protocol in local area network(LAN),this paper designs and implements an intrusion detection method of IPv6 LAN based on neural network.The feedforward neural network in deep learning is used to build the model,and the abnormal data stream caused by intrusion behavior is used as the original input to drive the training model.The system performance is optimized by combining the error back propagation and optimization function.Experiments on campus show that compared with the classical intrusion detection method based on pattern matching algorithm,the proposed method is significantly better than the algorithm based on pattern matching in terms of correct rate and false alarm rate,and under the same conditions,the number of detection is more than 5%than the algorithm based on pattern matching.

关 键 词：路由协议 局域网 入侵检测系统 网络性能 神经网络 

分 类 号：TN915.08[电子电信—通信与信息系统] TP393[电子电信—信息与通信工程]