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

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