- : Ms Word, Ms Word Format
- : 70 Pages
- : ₦3,000 | $25 | ₵60 | Ksh 2720
- : 1-5 Chapters
- Click to DOWNLOAD Materials
A HIGH-PERFORMANCE SYSTEM FOR ANONYMITY IN PEER-TO-PEER FILE TRANSFER NETWORKS
Abstract
In peer-to-peer file transfer, the properties of privacy and performance are often mutually exclusive. Although there are a wide variety of peer-to-peer applications and networks which attempt to close this gap, even the most elegant systems still present privacy and performance as start trade-offs to one another. Amidst continually increasing concerns about privacy, as well as ever growing amount of data being transferred, a peer-to-peer system which provides a more balanced approach to anonymity and performance is desperately needed. In this research, I propose a new system called Excelsior which embraces both of these properties in a hybrid network utilizing DC-nets, entanglement, and BitTorrent-style content distribution. This system is designed to provide a simple and straightforward means for the ubiquitous BitTorrent to vastly increase user privacy with the minimum possible performance sacrifice.
Table of Contents
List of Figures……………………………………………………………………………………………………………… viii
List of Tables………………………………………………………………………………………………………………… ix
Chapter 1: Introduction……………………………………………………………………………………………………. 1
1.1 Overview………………………………………………………………………………………………………………. 1
1.2 Research Objectives……………………………………………………………………………………………….. 4
Chapter 2: Literature Review……………………………………………………………………………………………. 5
2.1 Communication Models………………………………………………………………………………………….. 5
2.2 Peer-to-Peer Networking…………………………………………………………………………………………. 7
2.3 File Sharing…………………………………………………………………………………………………………. 10
2.4 BitTorrent……………………………………………………………………………………………………………. 12
2.5 Privacy and Anonymity: Defined……………………………………………………………………………. 14
2.6 Privacy and Anonymity: Trade-Offs……………………………………………………………………….. 17
2.7 Existing Solutions………………………………………………………………………………………………… 18
Chapter 3: Excelsior……………………………………………………………………………………………………… 20
3.1 Overview…………………………………………………………………………………………………………….. 20
3.2 Objectives for Resulting System…………………………………………………………………………….. 20
3.3 General Theory of Operation…………………………………………………………………………………. 21
3.4 Modes of Operation………………………………………………………………………………………………. 24
3.4.1 Overview………………………………………………………………………………………………………. 24
3.4.2 Single Source………………………………………………………………………………………………… 24
3.4.3 Multi-Source…………………………………………………………………………………………………. 25
3.5 Technical Description…………………………………………………………………………………………… 27
3.5.1 Common Elements…………………………………………………………………………………………. 27
3.5.1.1 Dining Cryptographers Network………………………………………………………………… 27
3.5.1.2 Block Identification and Validation……………………………………………………………. 30
3.5.1.3 Publishing and Advertisement…………………………………………………………………… 30
3.5.2 Single Source………………………………………………………………………………………………… 31
3.5.2.1 Entanglement Process………………………………………………………………………………. 31
3.5.2.2 Initial Seed-to-Seed Distribution……………………………………………………………….. 32
3.5.2.3 Client Downloads…………………………………………………………………………………….. 33
3.5.2.4 Reassembly…………………………………………………………………………………………….. 33
3.5.3 Multi-Source…………………………………………………………………………………………………. 34
3.5.3.1 Initial Seed-to-Seed Distribution……………………………………………………………….. 34
3.5.3.2 Entanglement Process………………………………………………………………………………. 35
3.5.3.3 Client Downloads…………………………………………………………………………………….. 35
3.5.3.4 Reassembly…………………………………………………………………………………………….. 36
Chapter 4: Performance Analysis……………………………………………………………………………………. 38
4.1 Overview and Methodology…………………………………………………………………………………… 38
4.2 Goals…………………………………………………………………………………………………………………… 39
4.3 Test Environment…………………………………………………………………………………………………. 39
4.3.1 Physical Hardware Configuration…………………………………………………………………….. 39
4.3.2 Virtual Machine Configuration………………………………………………………………………… 40
4.4 Metrics………………………………………………………………………………………………………………… 41
4.5 Test Configuration………………………………………………………………………………………………… 42
4.6 Test Results…………………………………………………………………………………………………………. 43
4.6.1 Phase Two Transfer Time……………………………………………………………………………….. 43
4.6.2 Overall Transfer Time…………………………………………………………………………………….. 45
4.6.3 Preparation and Reassembly……………………………………………………………………………. 49
Chapter 5: Security Analysis………………………………………………………………………………………….. 50
5.1 Overview and Methodology…………………………………………………………………………………… 50
5.2 Design Goals……………………………………………………………………………………………………….. 51
5.3 Components…………………………………………………………………………………………………………. 51
5.3.1 Matchmaking………………………………………………………………………………………………… 51
5.3.2 DC Nets and Entanglement……………………………………………………………………………… 52
5.3.3 Metadata Distribution…………………………………………………………………………………….. 53
5.3.4 BitTorrent and Reassembly……………………………………………………………………………… 53
5.4 Security Model…………………………………………………………………………………………………….. 53
5.4.1 Trust Model…………………………………………………………………………………………………… 53
5.4.2 Threat and Attack Model………………………………………………………………………………… 54
5.4.2.1 Privacy Compromise………………………………………………………………………………… 54
5.4.2.2: Content Alteration………………………………………………………………………………………….. 57
5.4.2.3 Network Disruption………………………………………………………………………………….. 59
5.4.3 Adversaries……………………………………………………………………………………………………. 60
Chapter 6: Conclusions………………………………………………………………………………………………….. 62
6.1: Overview……………………………………………………………………………………………………………. 62
6.2 Performance………………………………………………………………………………………………………… 62
6.3: Security……………………………………………………………………………………………………………… 63
6.4 Final Thoughts……………………………………………………………………………………………………… 64
6.5 Future Work………………………………………………………………………………………………………… 64
Appendix: Performance Benchmarks………………………………………………………………………………. 67
Bibliography………………………………………………………………………………………………………………… 73
List of Figures
Figure 2.1: Client/Server versus Peer-to-Peer ……………………………………………………….. 5
Figure 2.2: Client/Server Content Delivery …………………………………………………………… 8
Figure 2.3: Peer-to-Peer Content Delivery ……………………………………………………………10
Figure 2.4: BitTorrent Content Distribution …………………………………………………………13
Figure 3.1: DC Net Exchange ……………………………………………………………………………..28
Figure 4.1: Dell PowerEdge T605 Server ………………………………………………………………40
Figure 4.2: BitTorrent Network Download Time …………………………………………………..44
Figure 4.3: BitTorrent Network Overhead ……………………………………………………………45
Figure 4.4: Overall Download Time……………………………………………………………………..47
Figure 4.5: Overall Network Overhead …………………………………………………………………48
Figure 4.8: Preparation and Reassembly Times for Single Source Excelsior ……………….49
Figure 4.9: Preparation and Reassembly Times for Multi-Source Excelsior ……………….49
List of Tables
Table 2.1: Downstream Bandwidth Comparison ……………………………………………………. 9
Table 4.1: Performance Metrics …………………………………………………………………………..41
Table 4.2: BitTorrent over Tor Benchmarks …………………………………………………………43
Table 4.3: Excelsior BitTorrent Single Source Benchmarks ……………………………………..43
Table 4.4: Excelsior BitTorrent Single Source Benchmarks ……………………………………..43
Table 4.5: BitTorrent-over-Tor Overall Benchmarks ……………………………………………..46
Table 4.6: Excelsior Overall Single Source Benchmarks ………………………………………….46
Table 4.7: Excelsior Overall Single Source Benchmarks ………………………………………….46
Table A.1: BitTorrent over Tor ………………………………………………………………………….67
Table A.2: Single Source DC Net ………………………………………………………………………..68
Table A.3: Multi-Source DC Net …………………………………………………………………………69
Table A.4: Single Source BitTorrent ……………………………………………………………………70
Table A.5: Multi-Source BitTorrent …………………………………………………………………….71
Table A.6: Preparation and Reassembly ………………………………………………………………72
Chapter 1: Introduction
1.1 Overview
Peer-to-peer (P2P) communication is important. Spanning a wide variety of systems – everything from low-powered wireless sensor nets, to bandwidth-saving technologies in corporate networks such as Microsoft’s BranchCache[1] , to popular public file sharing software like BitTorrent – P2P communication is everywhere. As a mature technology, P2P is increasingly being used to facilitate the efficient transmission of large volumes of content.
The latter example above, BitTorrent, is a high-performance P2P file sharing protocol. BitTorrent accounts for a substantial percentage of global internet traffic, and is capable of disseminating a wide range of content, including but not limited to: software applications, audio, video, and textual data [4, 9, 16] . Although this protocol has gained popularity due to its efficiency and open-source nature, it lacks virtually any concept of user privacy or anonymity, as BitTorrent readily shares a large amount of information about a user, such as their IP address and what content they are uploading or downloading [1, 9, 13] .
A number of extensions or workarounds to the protocol have been proposed to address these shortcomings. One such example is routing BitTorrent traffic through anonymous overlay networks. Solutions like Tor and I2P provide a relatively safe haven, but at a significant cost in terms of bandwidth and latency [15] . In 2015, estimates place global BitTorrent bandwidth usage at around 1.8% of all traffic, or 4.97TB/s, while the total advertised bandwidth across the entire Tor network is only in the range of approximately 30GB/s [4, 16] . Limitations such as this prevent the wide-scale adoption of these measures.
Alternate approaches utilizing non-BitTorrent protocols, networks, and clients exist that provide the required user privacy/anonymity without the performance limitations that routing BitTorrent traffic through traditional overlay networks incurs, however these systems are generally standalone in that they do not communicate with other networks [15] . Since there is no benefit to using these systems if one does not require privacy or anonymity, it unfortunately becomes easy to label those users as “having something to hide”.
An ideal solution would protect user privacy/anonymity without sacrificing performance or functionality, all while allowing a hybrid network consisting of both standard and augmented peers [14] . Although this “silver bullet” is likely not a realistic goal, it is still possible to improve the current state of affairs by partially limiting the scope of the privacy or anonymity required to include only the elements deemed absolutely necessary to protect the user.
The approach to protecting user privacy/anonymity in a BitTorrent network is often the same as that used in other networks: an emphasis is placed on guaranteeing anonymity by preventing other users from associating in-network identities or activity with public identities, such as an IP address or name [6, 7, 10] . For some types of networks, this approach is appropriate, as the network exists for a singular purpose, which is the inferred purpose of any users of that network.
As an example of the above concept, one could generally infer that a user of an internet forum dedicated to mathematics is interested in learning about or discussing math. As an alternate example, the reputation of a user participating on a forum whose topic material was potentially illegal could be damaged simply by being a member of that forum.
BitTorrent networks are different, however. Although there are many types of content available, some of which is sensitive, there is plenty of content that is not [30] . For example, many open-source and freely available Linux distributions offer downloads through the BitTorrent network [30] . It is therefore not possible to accurately infer a user’s purpose simply by their participation, but rather by what the user does as a member of the network, and more specifically, what content they upload or download. This is an important distinction, as it allows for targeting the privacy measures in place to specifically address protecting the identity of data being transferred, rather than the much broader and more complex task of protecting the identity of a participant in said transfer.
1.2 Research Objectives
- To perform a survey of currently available methods for providing privacy/anonymity in BitTorrent networks.
- To devise a method by which to decrease the trade off between privacy/anonymity, and performance in BitTorrent networks.
- To develop an implementation of this method, conduct testing, and provide an analysis of the results.
[1] BranchCache is a trademark of Microsoft Corporation in the United States and/or other countries.
A HIGH-PERFORMANCE SYSTEM FOR ANONYMITY IN PEER-TO-PEER FILE TRANSFER NETWORKS
