Anonymity-Based Privacy Preserving Network Data Publication

Abstract

Network trace data provide valuable information which contributes to model the network behaviors, defend network attacks and develop new protocols, so releasing the data of network trace is highly demanded by researchers and organizations to promote the development of the network technologies. However, due to the sensitive nature of network trace data, it is a potential risk for organizations to publish the original data which may expose their commercial confidentiality and the customers' privacy within their networks. Several methods to defend the network trace attacks such as statistical fingerprinting and injection have been proposed, unfortunately, they are not enough to protect the privacy because the correspondence between the source and destination IP addresses can also help the adversary to identify the target host. In this paper, we extract the inherent graph structure between the source and destination IP addresses in network trace data, and use k-anonymity to prevent the target host from being identified. Combined with other protection techniques, our method can also prevent the fingerprinting and injection attacks.