Puzzles in P2P Systems - PowerPoint PPT Presentation

1 / 14
About This Presentation
Title:

Puzzles in P2P Systems

Description:

Here we look at P2P 'stuff-sharing' systems. Kazaa, Gnutella, etc, etc, etc. Bit Torrent. Abstract from the politics. RIAA. ISP's. Motivation. The Free Riding Problem ... – PowerPoint PPT presentation

Number of Views:43
Avg rating:3.0/5.0
Slides: 15
Provided by: gtad7
Category:

less

Transcript and Presenter's Notes

Title: Puzzles in P2P Systems


1
Puzzles in P2P Systems
  • Andrei Serjantov and Stephen Lewis
  • University of Cambridge Computer Laboratory

2
Introduction
  • Here we look at P2P stuff-sharing systems
  • Kazaa, Gnutella, etc, etc, etc
  • Bit Torrent
  • Abstract from the politics
  • RIAA
  • ISPs

3
Motivation
  • The Free Riding Problem
  • Downloading without sharing
  • The Slashdot Effect
  • MojoNation
  • Performing Community Tasks
  • Quality Rating
  • Indexing

4
Sketch Solution
  • Protocol
  • When user asks for content, hand them a puzzle
  • They solve the puzzle, send you the solution
  • You send them the content
  • Both are happy
  • What is the puzzle?
  • Who generates it? How?
  • Different answers help provide different
    properties
  • Serious thought needed

5
An Example Protocol
B
Yep, Solve The Halting Problem!
Puzzle Server
LOTR?
Halting problem?
A
Here it is!
6
Solving the Halting Problem
A
Hmmmmmmm Hmmmmmmm Hmmmmmmm Hmmmmmmm Halts!
7
Exmaple Protocol (cont)
B
Halts!
LOTR III
A
8
Encouraging Content Sharing
  • Assume a central puzzle repository
  • Much more stuff than puzzles
  • When a piece of stuff is requested, the to be
    solved puzzle is specified (as above)
  • A user who shares lots of stuff has many puzzle
    solutions, hence does not have to solve often
    when downloading
  • Periodically, a fraction of the puzzles in the
    repository is changed

9
What is a Puzzle?
  • Computational puzzle
  • Processor-based computation
  • Hashcash
  • Memory-bound
  • Community puzzle
  • Rating content quality
  • Indexing
  • Turing puzzle
  • Humans can do and computers cannot
  • See below

10
Computational Puzzles
  • Processor puzzles
  • Finding a hash function collision
  • Easy to check
  • Memory-bound computational puzzles
  • Used for anti-spam schemes
  • How fast you solve the puzzle depends on the
    amount of memory, not (so much) on processor speed

11
Community Puzzles
  • Essentially as above, but the user has to perform
    some task which benefits the entire community at
    the cost of his convenience
  • Quality rating
  • Indexing
  • Verifiable
  • Majority voting will do
  • Subset of Turing puzzles

12
Money Money Money
  • Micropayments as puzzles ??
  • Popularity tends to follow the power law
    distribution
  • Hence, collecting a small payment from many will
    enable a good payoff to the popular few.
  • (Do you care if your subscription charge is /- 1
    Euro?)
  • Micropayments are pure speculation
  • More likely to succeed in the context of a single
    system

13
Research
  • Investigate security protocols with entirely new
    properties
  • non-cheating
  • equal benefit
  • Rather than secrecy or authenticity
  • Some properties will be ensured not by
    cryptographic means, but by rationality of the
    users
  • Information Security Economics
  • Which puzzle are best?

14
What I really do(!)
  • Censorship Resistance
  • Privacy
  • Mostly politics(!)
  • Information Security Economics
  • Anonymity
  • Web browsing
  • Email
  • P2P?
  • P2P

See http//www.cl.cam.ac.uk/aas23/
Write a Comment
User Comments (0)
About PowerShow.com