Making Peer-Assisted Content Distribution Robust to Collusion Using Bandwidth Puzzles
Source: University of North Carolina
Many peer-assisted content-distribution systems reward a peer based on the amount of data that this peer serves to others. However, validating that a peer did so is, to the authors' knowledge, an open problem; e.g., a group of colluding attackers can earn rewards by claiming to have served content to one another, when they have not. They propose a puzzle mechanism to make contribution-aware peer-assisted content distribution robust to such collusion. Their construction ties solving the puzzle to possession of specific content and, by issuing puzzle challenges simultaneously to all parties claiming to have that content, their mechanism prevents one content-holder from solving many others' puzzles.