自由空间可重构光互连网络入侵数据检测技术  

Intrusion detection technology for reconfigurable optical interconnection network in free space

作　　者：王维 易禾 

机构地区：[1]四川师范大学信息网络服务中心

出　　处：《激光杂志》2018年第1期133-136,共4页Laser Journal

基　　金：四川省教育厅科研项目(No.SCYG2015-2-12)

摘　　要：自由空间可重构光互连网络传输数据量大,非法入侵频率高,针对传统入侵检测技术存在的不足,提出一种主动学习和半监督聚类的光互连网络入侵数据检测技术。首先分析了自由空间可重构光互连网络的工作原理,对光互连网络数据进行预处理,然后利用有标记样本训练产生2个分类器,通过分类器完成对没有标记样本的预测,将结果不同的样本看作熵值最大的样本,通过新得到的有标记样本集继续对主动学习和半监督分类器进行训练,直至没有标记样本集为空,最后通过具体实验结果对其性进行测试,结果,该技术提高了网络入侵检测的精度,可以有效保证自由空间可重构光互连网络安全性。Free space of reconfigurable optical interconnection network has transmitting data, illegal intrusion of high frequency, aiming at the shortage of the traditional intrusion detection technology, this paper proposes a new opti- cal interconneetion network intrusion data detection technology for active-learning and semi-supervised clustering. Firstly, analyzes the principle of reconstruction of free space optical interconnection network, optical interconnection network data are pretreated,. Secondly, the training labeled samples are used to produce two classifiers to predict the completion of no labeled samples. The results of different samples are taken as the maximum entropy sample. And con- tinue to train active-learning and semi-supervised classifier through new labeled samples, until no labeled sample set is empty. Finally tests functions through concrete experimental results, the results show that this technology improves the network intrusion detection accuracy, and can effectively guarantee security.

关 键 词：自由空间 可重构 光互连网络 入侵数据 检测 

分 类 号：TN256[电子电信—物理电子学]