The rest of the paper is organized as follows. Section 2 reviews the existing P2P
anonymous routing schemes. Our approach is presented in Section 3. We present the
anonymity analysis of our approach in Section 4. In section 5, we employ the
simulation to demonstrate the effectiveness of our approach. We conclude the paper
with future research directions in Section 6.
2 Related Work
Most anonymous routing protocols can be categorized into mix-based and multicast-
based types. Mix-based systems [Chaum 1981] achieve anonymity by applying the
redirection technique such as Tor [Dingledine et al. 2004], while multicast-based
systems achieve anonymity by employing multicasting technique such as P5
[Sherwood et al. 2005], Hordes [Shields et al. 2000] and APFS [Scarlata et al. 2001].
As P2P networks are becoming appealing platforms for constructing anonymous
systems, a number of P2P-based anonymous systems have been proposed. Tarzan
[Freedman et al. 2002] and MorphMix [Rennhard et al. 2002] achieve anonymity by
applying layered encryption and multi-hop routing approaches. In contrast, Crowds
[Reiter et al. 1998] achieves anonymity by applying probabilistic random forwarding
scheme. Although P2P-based anonymous systems aim to achieve more anonymity,
the performance is degraded due to the node churn in P2P networks, which makes
anonymous paths fragile and short-lived.
To improve the performance of P2P networks, several incentive mechanisms
have been proposed. [Sun et al. 2004] proposed a simple Selfish Link-based
InCentive (SLIC) approach for unstructured P2P file sharing systems where nodes in
exchange for better services are encouraged to share more data, give more capacity to
neighbor nodes’ queries, and add new overlay links. To rate neighbor nodes
efficiently, several incentive mechanisms have been proposed recently. [Barth et al.
2008] proposed a complete distributed solution to the transit price negotiation
problem with incomplete information. [Wu et al. 2008] proposed the approach to rank
retrieval systems in the condition of partial relevance judgments. Such incentive
mechanisms need to be studied further for being employed in anonymous network.
To make anonymous paths resilient to node failures, TAP [Zhu et al. 2004] and
Cashmere [Zhuang et al. 2005] decouple paths from fixed relay nodes and use a group
of nodes in structured P2P networks [Rowstron et al. 2001] to mask single relay node
failures by sharing group keys among group members. The main limitation of such
systems is that group members must share some secrecy such as group keys. In order
to avoid such limitation, the re-encryption mechanism combined with key exchange
protocols can be used. [Ateniese et al. 2006] proposed an improved proxy re-
encryption scheme allowing third-parties (proxies) to alter a ciphertext which has
been encrypted for one party to be decrypted later by another. [Jeong et al. 2008]
proposed an efficient parallel key exchange protocol among multiple parties and
[Hwang et al. 2008] proposed the secure CL-PKE scheme to withstand malicious key
generation centre attack.
MuON [Bansod et al. 2008] is a mutual anonymity system that uses epidemic-
style data dissemination to handle network dynamics in unstructured P2P networks.
Zhu et al. [Zhu et al. 2007] proposed a message redundancy scheme to mask node
failures by employing erasure coding and path redundancy approaches. Although such
1799
Luo J., Wang X., Yang M.: A Resilient P2P Anonymous Routing Approach ...