Nodes detection in CDN with enhanced naive bayes tree
Ruijjie Hou, Changyong Guo, Zhaoxin Zhang, Ning Li · Journal of Physics Conference Series · 2021
Abstract In recent years, Content Distribution Network (CDN) plays a critical and central part of Internet infrastructure. There are more and more studies on CDN. Nodes detection in CDN, as a key technology among them, has become a hot research topic. Current methods mainly focus on collecting one or some CDN vendors nodes by manually constructing features set. However, because CDN nodes detection is not restricted by specific CDN vendors, these methods are not applicable. In this paper, we proposed a novel machine learning algorithm, i.e. enhanced Naive Bayes Tree to identify CDN IP addresses. This algorithm makes full use of the advantages of Decision Tree and Naive Bayes algorithm, and further improves the performance through enhanced part. We build this classifier based on analysis of the characteristics of DNS resolutions, HTTP logs and WHOIS lookup. Experimental results show that our approach outperforms in terms of accuracy and amount of packages. Moreover, we separately tested the enhanced part and it performs well. We believe that this method could be applied to CDN IP addresses detection.