Anonymizing Network Addresses Based on Clustering Subnets

The network trace is a kind of fundamental data for networking researches. To preserve the privacy hidden in those traces, they must be sanitized before publishing them publicly. Many of the sanitization efforts are focused on anonymizing internal network addresses. In this paper, we propose a subnet-clustering based method to anonymize addresses. We adopt different strategies to anonymize different parts of the address. The network part is anonymized by a prefix-preserved anonymization method, the subnet part is generalized by clustering based on a predefined set of port numbers, and the host address is randomized. We also propose a measure based on information entropy to measure the degree of privacy-preserved in anonymized addresses and develop an entropy-guided algorithm to search the subnet clusters.