An effective index poisoning algorithm for controlling peer-to-peer network applications

  • Authors:
  • Pratama Putra;Akihiro Nakao

  • Affiliations:
  • The University of Tokyo, Hongo, Bunkyo, Tokyo, Japan;The University of Tokyo, Hongo, Bunkyo, Tokyo, Japan

  • Venue:
  • Proceedings of the 2011 International Workshop on Modeling, Analysis, and Control of Complex Networks
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Copyright infringement is considered a significant issue in P2P network communications. Index and content poisoning have been proposed to control the exchange of copyrighted content. Unfortunately, however, such control methods are costly in terms of the amount of control traffic since they apply the control method to all the peers and they generate much control traffic. In general, directly applying index poisoning to a peer may indirectly poison the neighboring peers, thus, it is possible to reduce the number of peers to target for the same effect of poisoning. In this paper, we propose a method to improve index poisoning by limiting the scope of poisoning so that even when we apply poisoning to a small number of peers, we could still achieve the same effect of traffic control as we applied poisoning to all peers. In more detail, taking Winny as an example of structured P2P network, we propose, implement and evaluate an algorithm to determine the influential peers for index poisoning based on the inferred network structure. We successfully reduce the poisoning target to 27% of the total peers and achieve the same effectiveness as the conventional method poisoning 96% of the entire network.