Overview

Nowadays anonymous networks impose a trade-off between anonymity and traffic performance. A high level of anonymity can be achieved primarily at a high traffic expense. In particular, the latency constitutes a crucial factor for performance of anonymous networks, as the basic common mechanism used to achieve network anonymization is the traffic forwarding by a set of middleman nodes (also known as proxy chaining).

The new idea behind P2PRIV can be stressed as the Parallel Anonymization.

P2PRIV introduces the parallel content transport instead of the widespread cascade transmission via a chain of proxies. Certainly, anonymity assured by the P2PRIV imposes traffic overheads, as in any other anonymous system. A motivation behind the parallel architecture with the direct content transport is a decrease of the download time while preserving high degree of anonymity. In the P2PRIV only short control messages are sent over classical anonymous cascades, called Cloning Cascades (CC). In brief, the P2PRIV uses the well known anonymous techniques (random walk algorithm and digital mixes chaining) for anonymization of a specific management communications adjusted to provide further anonymous and direct parallel transport of the shared information content.

The two steps of P2PRIV operations are:

• Step 1: Cloning – an anonymous and random selection of a subset of nodes referred to as the Cloning Cascade. Each such CC contains the requestor and its clones; each peer can be potentially selected for such a clone. The initiator sends a token with a file id to a randomly chosen peer. Then, the selected peer flips an asymmetric coin to decide whether to forward the token to the next random peer. This communication may be additionally secured and anonymized by mix-net mechanisms, as numerous but short control messages of constant length, generated by cloning, can be effectively exchanged by the mix cascades.
• Step 2: Data connection – transport of the requested content. After a random interval of time and based on the content id received earlier, the copies of the content are directly downloaded by selected (cloned) peers from nodes which store data. Files can be looked up by the DHT algorithm. As in the cese of the cloning exchange, look-up messages can be effectively secured by mix-net mechanisms. The resulting data redundancy is similar to classical solutions. It improves content accessibility, because the popularity of a content automatically increases the number of its copies stored in an overlay network.

Notice, that in the solution the anonymization process is separated from the transport of information content, in contrast to classical schemes. We believe this can facilitate high-anonymous—low-latency P2P communications.

To learn more about P2PRIV, please take a look at P2PRIV's Documents.