论文部分内容阅读
随着跳板主机和匿名网络成为隐匿通信关系的常用手段,网络攻击流量的溯源和定位难度日益增大.网络流水印技术在网络隐私安全领域已逐渐成为了一种重要的网络流量追踪和定位手段,设计良好的网络流水印具有强大的鲁棒性和隐蔽性,使得对网络流水印的存在性检测变得异常困难,而对流水印实施有效检测是进一步实现水印移除或水印流量复制的前提.本文提出了一种基于多流联合质心熵的水印盲检测方法,其可以实现针对当前典型的时隙质心类流水印的有效检测.在实际SSH流量上的实验结果表明,所提方法在单密钥情形下可达到与当前普遍采用的多流攻击相近的检测效果,在随机多密钥情形下多流攻击方案失效而本文方案依然可以实现高效检测.
As springboard hosts and anonymous networks become a commonly used means to conceal communications, traceability and positioning of network attack traffic are increasingly difficult.Network watermarking technology has gradually become an important network traffic tracing and positioning means in the field of network privacy security , The well-designed network watermarking has strong robustness and concealment, making it difficult to detect the existence of network watermarking, and the effective detection of the watermarking is the prerequisite for further watermark removal or watermarking traffic replication. This paper proposes a blind watermark detection method based on multi-stream joint centroid entropy, which can effectively detect the current typical class of centroid watermarks.Experimental results on the actual SSH traffic show that the proposed method in single-dense Key case, it can achieve the detection effect similar to the multi-stream attack commonly used today, and the multi-stream attack solution fails in the case of random multi-key and the scheme in this paper can still achieve efficient detection.