Why does Tor reject loss-based congestion control algorithms? Do I understand correctly that the main idea is that in this case 1 hostile node can intentionally drop packets, causing changes in the traffic pattern on the sender’s side? But when using Delay-based algorithms (such as tcp vegas, on which the current TOR algorithm is based), a hostile node can also add artificial delays, causing changes in the traffic pattern). So why is one class of algorithms rejected, and the second accepted?
1 Like