Network covert channel analysis based on the density multilevel two segment clustering

On the problem of covert channel detection, the traditional detection algorithms exist specific covert channel blind area, or it is useful for some kind of covert channel detection but ignore other covert channels. In order to solve this problem, in this paper proposes network covert channel analysis method based on the density multilevel two segment clustering. Firstly, the problem of covert channel in complex network is studied, and its mathematical model and data feature extraction are presented; Secondly, based on hierarchical clustering and design its multilevel aggregation improved form using the given complex network channel coarsening clustering results, at the same time in each layer of coarse channel and the results of detection, using density clustering algorithm to implement complex network covert channel detection and thinning and improve the prediction accuracy. Finally, the proposed algorithm can detect the complex network covert channel quickly and accurately when the noise is no higher than 20%.