基于粗糙集的攻击意图特征提取研究

发布时间：2018-05-13 12:15

本文选题：粗糙集 + 属性约简　；参考：《河北师范大学》2017年硕士论文

【摘要】：在当今大数据时代的背景下,随着数据种类增多、规模扩大,大数据在各个领域中的重要性也与日俱增,安全事件的层出不穷使大数据安全问题面临严峻的考验。网络安全技术一直在不断变革,从传统的入侵阻止、入侵检测发展到入侵容忍、可生存性研究,从关注信息的保密性发展到关注信息的可用性和服务的可持续性,从关注单个安全问题的解决发展到研究网络的整体安全状况及变化趋势,网络安全态势评估已成为网络安全技术的焦点。而态势要素的提取是网络安全态势评估的基础,包括网络结构、资产、威胁等方面。前两方面信息是基本固定的,威胁是未知的不确定的,而攻击意图又是威胁中的关键所在。因此攻击意图的提取也变得尤为重要。然而现有的攻击意图提取方法由于数据来源存在海量、冗余及错误等缺陷,对攻击意图的深层次分析不足,多源异构数据背景的攻击意图提取技术不成熟等原因已无法满足当前网络安全的需求。基于以上原因,本文提出了运用粗糙集对攻击意图特征进行提取的方法。基于粗糙集的攻击意图特征提取方法:首先对数据集进行预处理。删除特殊属性与攻击,以达到降维的目的,然后对数据库根据适当的属性进行分块,以简化数据集,更好地优化属性得到更理想的结果。进而在简化的数据集基础上对其进行离散化,根据数据集特征选用合适的离散算法,并与原始数据集离散效果进行对比。基于离散化结果,对数据集属性特征进行约简,运用提出的基于改进的回溯搜索优化算法的决策粗糙集约简方法提取其重要特征,以达到相对较好的结果,为攻击意图的预测提供有效的数据基础。
[Abstract]:Under the background of today's big data era, with the increase of data types and the expansion of the scale, the importance of big data in various fields is also increasing day by day. The emergence of security incidents in endlessly makes the big data security problems face a severe test. Network security technology has been changing, from traditional intrusion prevention, intrusion detection to intrusion tolerance, survivability research, from information confidentiality to information availability and service sustainability. From focusing on the solution of individual security problems to studying the overall security situation and changing trend of the network, the network security situation assessment has become the focus of the network security technology. The extraction of situation elements is the basis of network security situation assessment, including network structure, assets, threats and so on. The first two aspects of the information are basically fixed, the threat is unknown and uncertain, and the intention of attack is the key to the threat. Therefore, the extraction of attack intention becomes particularly important. However, due to the defects of data sources, redundancy, error and so on, the existing attack intention extraction methods are insufficient in the deep level analysis of the attack intention. The attack intention extraction technology of multi-source heterogeneous data background is not mature, and so on, which can not meet the current network security requirements. Based on the above reasons, a rough set method is proposed to extract the feature of attack intention. The feature extraction method of attack intention based on rough set: firstly, the data set is preprocessed. The special attributes and attacks are deleted to achieve the purpose of dimension reduction, and then the database is divided into blocks according to the appropriate attributes to simplify the data set and better optimize the attributes to obtain more ideal results. On the basis of the simplified data set, the discrete algorithm is selected according to the feature of the data set, and compared with the result of the original data set discretization. Based on the discretization result, the attribute feature of the data set is reduced, and the decision rough set reduction method based on the improved backtracking search optimization algorithm is used to extract its important features, so as to achieve a relatively good result. It provides an effective data base for the prediction of attack intention.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TP393.08

【相似文献】