摘要
网络入侵检测必须面对海量的数据获取和处理,压缩感知理论能够直接、快速压缩采集网络中的数据流.提出一种基于压缩感知的入侵检测方法,该方法通过对访问数据的压缩采样,获取正常和异常行为的特征数据.这种数据处理方式避开了大量的数据处理,直接获取特征,这对于网络入侵检测需要进行高维的数据处理过程来说,大大节省了处理时间,为实现实时的入侵检测提供了重要的技术手段.
Network intrusion detection often confronts the challenge of the acquisition and processing of da ta with huge volumes. Compressed sensing makes it possible to directly and rapidly compress the dala stream collected from network. The present paper proposes an intrusion detection technique based oncom pressed sensing that is able to acquire both normal and abnormal feature data through compressive sam pling on the access data. This technique effectively avoids processing large amount of data by means of di rect acquisition of the feature. As far as high-dimensional data processing is concerned, this method signif icantly saves the processing time and serves as an important means to achieve real time intrusion detection.
出处
《西南大学学报(自然科学版)》
CAS
CSCD
北大核心
2014年第2期186-192,共7页
Journal of Southwest University(Natural Science Edition)
基金
国家自然基金资助项目(61170192)
中央高校基本科研业务费专项资金资助(CDJXS11180007
XDJK2120131037)
关键词
入侵检测
压缩感知
重构
稀疏采样
intrusion detection: compressed sensing
reconstruction
sparse sampling