UC San Diego

Tor is an onion routing network that protects users' privacy by relaying traffic through a series of nodes that run Tor software. As a consequence of the anonymity that it provides, Tor is used for many purposes on the internet including interactive traffic as well as for bulk file downloads. Such bulk downloads cause delays for interactive traffic as all traffic between a pair of Tor nodes goes over a single connection. The resulting delays discourage people from using Tor for normal web activity. We propose a potential solution to this problem called Torchestra which separates interactive and bulk traffic onto two separate TCP connections between any pair of nodes. We classify a circuit as carrying either type of traffic based on the Exponentially Weighted Moving Average of its number of cells. We evaluate our proposal by simulating traffic using several methods and show that Torchestra provides up to 32% reduction in delays for interactive traffic compared to the Tor traffic prioritization scheme of Tang and Goldberg and up to 40% decrease in delays when compared to unprioritized Tor