Tor official page says “Tor fails when the attacker can see both ends of the communications channel.”

Then,setting up a bridge on pc A and using it myself as entry node on pc B (with vpn to change ip)can make this attack impossible? (Because attacker can never be the entry node in this case.)

Or, asking reliable friends to be the bridge each other is also effective to improve anonymity?

  • No, the attacker needn't be the entry, just see the traffic. – cacahuatl Jan 7 '17 at 9:17
  • I guess in fairnes this would prevent some attackers. I can't find this answered elsewhere, so I'll try to formulate and write a proper answer but tl;dr - won't beat adversaries who are watching you and can see your exit traffic. In the adversaries it might beat, a bridge may not be the best option. – cacahuatl Jan 7 '17 at 12:05

Yes, having friends running private/unpublished bridge is a way to improve your anonymity, but here is a caveat : use a pluggable transport like obfs3/4 and use bridge key you obtain/give offline, i.e. in personal meeting. It helped in some cases when other things were failing, even a single-hop circuits with pre-seeded tor haven't worked, but this setup did the trick. The key is a pluggable transport with the bridge key, other cases exposed Tor protocol to the bridge and failed. If your friends have a static IP's and you have already tried without a plug+key - be warned, some systems have banned the IP and rising a transport afterwards haven't worked so far

If you want to use just bridges to improve anonymity - no, it won't change the de-anonymization difficulty class from the case you're just using Tor without them. Bridges are for making it work when it is not working using default setup.

  • Censorship of the bridge has nothing to do with anonymity. Can you explain why your initial assertion is correct in your answer? – cacahuatl Jan 7 '17 at 21:03
  • Well, tor is an anonymity tool, so using it improves anonymity(that's what the Tor is made for). I had similar experiment with my lectures attendees in 2014: in one place was a very intrusive/restrictive firewall, but using home-based bridges with static IP's and pluggable transports with keys did the trick. The system has a protocol detector: using bridges without obfs has worked, but not so good. Using bridge without a key(it was a mistake) revealed the scheme described in answer. – Alexey Vesnin Jan 7 '17 at 21:11
  • 1
    No, can you explain why a private bridge provides better anonymity than normal Tor in the attack outlined in the question? – cacahuatl Jan 7 '17 at 21:41
  • 1
    Both ends means your connection and the exit point. To reach the bridge you'd have to traverse the adversary controlled network, Q.E.D. they see both ends of the connection. – cacahuatl Jan 7 '17 at 23:09
  • 1
    well, if an active adversary is able to see client, bridge(entry) and exit - it can just drop connections, so it won't work. My approach worked when bridge was out of the adversary control. – Alexey Vesnin Jan 7 '17 at 23:11

If the adversary can see both your connection into the network and the results of your connection then it's likely that they will, given time, be able to deanonymize you with some level of certainty. This is a fundamental limitation of all (known, at least) low latency anonymity networks.

Picking a known-good entry point will defend you in cases where the adversary doesn't have good network visibility of large enough portions of the network. In that case they could try running a set of entry nodes and some other point of interest (exit, hsdir, etc). By doing this, you'd remove their ability to link the traffic back to you. They might be able to control the "exit" and the middle node but since they don't have network visibility they'd have no means of linking it back to you directly, at best back to your known-good entry point...but there's the rub.

In your stated example of either running a private bridge yourself or having someone you known to run the private bridge, given that an adversary was able to track it back to the entry point, the entry point itself would have a restrictively small set of users. Thus, the fact that you are using the private bridge would reveal information about who you are, at least reducing it to a small set of possibilities.

To this end, if you have a known-good guard then you'd have a better anonymity set in the case where they were able to successfully discover your chosen entry point. However picking your own guard comes with a whole other set of potential pitfalls.

However, in all cases the first point still applied: if they can see enough of the network then you will still lose.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.