Approach to Recognition of Malicious Servers of TOR Anonymization Network Based on Cluster Analysis

Problems of secure communications on the Internet and anonymous access to network resources are considered. A problem is revealed, which involves an increase in the probability of deanonymization of TOR network users when using the servers under common administrative control in one chain. An approach to recognition of “hidden groups” of servers of TOR anonymization network through analysis of the server characteristics frequency and their clustering based on similarity is suggested. These studies demonstrate the approaches to improve software of the TOR anonymization network and increase user security.