A Sampling Method for Volume Entropy Analysis
Juan Wang, Peng Jing · International Journal of Digital Content Technology and its Applications · 2011
Entropy-based approaches for anomaly detection are appealing since they provide more finegrained insights than traditional traffic volume analysis. However, there has been little effort to make it more efficient on the large scale and high speed network, especially for the volume entropy analysis. The sampling methods of volume entropy analysis are studied in this paper. First of all, the sampling constraint of volume entropy analysis is studied and the heavy-tailed distribution of Netflow records is released. Then a special stratified sampling named record-size based regression sampling is proposed. The experiments show it observably decreases the processing time and storage space without losing much fluctuation accuracy.