Unmasking Tor users with timing signals
from CyberPodcast Hosted by CyberPodcast
This episode explores a powerful evolution in network traffic analysis, based on the paper "A Practical Congestion Attack on Tor Using Long Paths". While the original "congestion attacks" discovered in 2005 were once capable of unmasking Tor users, the podcast explains how they eventually became obsolete; as Tor grew to thousands of active relays, the network simply became too heavily loaded and noisy for a basic attacker's interference to reliably stand out. Listeners will discover how researchers resurrected this threat by exploiting a critical design flaw: at the time, Tor allowed users to build routing paths of arbitrary length. The episode breaks down an ingenious "bandwidth amplification" technique. Instead of using massive external server farms to overwhelm a target, an attacker can build a single, extremely long circuit that intentionally loops back through the same specific Tor router over and over again. By pumping just a small amount of data through this looped path, the attacker forces the target node to process the same traffic multiple times, artificially creating severe local congestion while using very little of their own bandwidth. The podcast then reveals how this loop is weaponized to systematically deanonymize a user. The attack begins when a malicious Tor exit node silently injects a tiny piece of JavaScript into the victim's web browser, forcing it to send a ping to the attacker every single second. By observing the precise arrival times of these pings while simultaneously unleashing their looping congestion attack on different suspected entry relays, the attacker essentially acts like a digital sonar. When the victim's ping latency suddenly spikes in perfect synchronization with the attacker's localized congestion, the victim's critical Guard node is definitively unmasked, successfully tracing the user's hidden path.
Embed this episode
No transcript yet.