On the Perfect Privacy: A Statistical Analysis of Network Traffic Approach

摘要

Many recent studies reveal that merely encrypting the source content does not inhibit the adversary from gaining knowledge about the source’s behavior. Thus, the adversary might be able to extract information from the network traffic by employing statistical analysis. Based on the information theory, in this letter, we quantify the amount of knowledge obtained by an adversary that overhears the channel. We present an approach that mixes the features of applications in the source node, such that it maximizes the ambiguity of adversary. Finally, we suggest three lower bounds of adversary’s error probability.