


摘 要:为了保障网络系统的安全与稳定,需要进行网络入侵检测模型研究,目前具有代表性的检测系统是基于遗传算法找出网络入侵的特征子集,但该系统检测准确性较低且训练时间过长。为此,本文将特征选择算法应用到网络实时入侵检测系统中,提出了一种基于特征选择的实时入侵检测方法。通过搭建非法入侵检测实验平台将该方法与基于遗传算法的网络入侵检测方法做比较,实验结果表明,该方法在检测攻击的准确率方面优于另一入侵检测系统,并且所需检测时间也短于另一检测系统。
关键词:特征选择算法;网络入侵;实时检测系统
中图分类号:TP393.08;TP181 文献标识码:A 文章编号:2096-4706(2019)20-0157-03
Abstract:In order to ensure the security and stability of network system,it is necessary to study the network intrusion detection model. At present,the representative detection system is based on genetic algorithm to find the characteristic subset of network intrusion,but the detection accuracy of the system is low and the training time is too long. Therefore,feature selection algorithm is applied to network real-time intrusion detection system,and a real-time intrusion detection method based on feature selection is proposed. This method is compared with the network intrusion detection method based on genetic algorithm by building an experimental platform of illegal intrusion detection. The experimental results show that this method is superior to the other intrusion detection system in detection accuracy,and the detection time is shorter than the other detection system.
Keywords:feature selection algorithm;network intrusion;real-time detection system
0 引 言
特征选择算法是近几年来发展起来的一类安全算法,它通过分析挖掘、检测网络中的数据流量的相关性、关联性,已经广泛应用在金融,建筑,信息安全等领域[1],从中发现网络信息系统中是否有违反安全策略的关联网络和被破坏置信度的迹象。它弥补了神经网络模式的防护措施缺口。随着攻击类型越来越多,蠕虫、木马、漏洞攻击、逻辑炸弹等威胁通常混合出现,留给系统恢复的时间越来越短,网络安全受到新的挑战[2]。原始的入侵检测机制通过模拟自然界的进化机制,如遗传信息寻优编译机制,找到粒子群数据特征集,取得了不错的检测效果。但是遗传算法要求样本大、成本高,学习速度慢。无法把攻击防御在企业网络之外,影响入侵检测的实时性,因此网络入侵检测具有广泛的发展空间。……