Publicação
TIRDEANON. DEANONYMIZATION AND TRAFFIC-UNOBSERVABILITY ANALYSIS OF TIR AS A STRENGTHENED SOLUTION TO IMPROVE ANONYMITY GUARANTEES IN TOR
| Resumo: | Tor is the most popular anonymity network used by Internet users. It is based on the core notion of multipath onion-encrypted routing and protected hidden services for anonymous access. Unfortunately, as shown in the recent research, Tor is vulnerable to deanonymization threats and censorship activities that can be conducted with high accuracy by state-level and global-collaborative entities cooperating as adversaries. TIR is a research solution designed to improve the anonymization guarantees for Tor users. TIR was targeted as an intermediary internetworking environment supported by collaborative ad-hoc community network nodes organized by a group of Tor users. A TIR network can be used as a pre-staged network for Tor, using K-anonymity input circuits protected as TLS-tunneled traffic. By using TIR for strengthening Tor, the original traffic from specific users is decoupled from the ingress traffic on well-known Tor entry relay nodes or Tor bridges. The protection is provided through an overlaying multipath and configurable traffic fragmentation strategy supported by the cooperation of K-anonymization internetworked nodes. In this dissertation, we study different techniques to deanonymize traffic flows with possible breaks of privacy-preservation of Tor traffic. Our approach leverages on the related work and recent results showing how traffic correlation based on machine learning techniques and deep neural networks can deanonymize Tor traffic. Based on the conducted study, we propose a system targeted as a tool and methodology for traffic observability analysis. The proposed solution extends the experimental assessment and analysis of unobservability criteria provided by TIR as an effective candidate solution to strengthen Tor users with stronger privacy-preservation arguments. In these arguments, we include resistance against active watermarking traffic correlations or possible privacy breaks of the K-anonymity sets of TIR-protected users. |
|---|---|
| Autores principais: | Carvalho, Bruno Miguel de Oliveira |
| Assunto: | Internet censorship Anonymization networks Tor Network Torenforcement with K-anonymized input circuits Covert circuits Tor traffic correlation attacks |
| Ano: | 2023 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | Tor is the most popular anonymity network used by Internet users. It is based on the core notion of multipath onion-encrypted routing and protected hidden services for anonymous access. Unfortunately, as shown in the recent research, Tor is vulnerable to deanonymization threats and censorship activities that can be conducted with high accuracy by state-level and global-collaborative entities cooperating as adversaries. TIR is a research solution designed to improve the anonymization guarantees for Tor users. TIR was targeted as an intermediary internetworking environment supported by collaborative ad-hoc community network nodes organized by a group of Tor users. A TIR network can be used as a pre-staged network for Tor, using K-anonymity input circuits protected as TLS-tunneled traffic. By using TIR for strengthening Tor, the original traffic from specific users is decoupled from the ingress traffic on well-known Tor entry relay nodes or Tor bridges. The protection is provided through an overlaying multipath and configurable traffic fragmentation strategy supported by the cooperation of K-anonymization internetworked nodes. In this dissertation, we study different techniques to deanonymize traffic flows with possible breaks of privacy-preservation of Tor traffic. Our approach leverages on the related work and recent results showing how traffic correlation based on machine learning techniques and deep neural networks can deanonymize Tor traffic. Based on the conducted study, we propose a system targeted as a tool and methodology for traffic observability analysis. The proposed solution extends the experimental assessment and analysis of unobservability criteria provided by TIR as an effective candidate solution to strengthen Tor users with stronger privacy-preservation arguments. In these arguments, we include resistance against active watermarking traffic correlations or possible privacy breaks of the K-anonymity sets of TIR-protected users. |
|---|