Reputation Systems for Self-Organized Networks

Abstract

Self-organized networks such as mobile ad-hoc, Internet-based peer-to-peer, wireless mesh and Fourth generation (4G) wireless networks depend on cooperation of nodes. Reputation systems help nodes decide with whom to cooperate and which nodes to avoid. They have been studied and applied almost separately in diverse disciplines such as economics, computer science, and social science, resulting in effort duplication and inconsistent terminology. We aim to bring together these efforts by outlining features and fundamental questions common to reputation systems in general. We derive methodologies to address these questions for both reputation system design and research from our own experiences and evaluations by simulation and analytical modeling. We argue for using deviation tests, discounting, passing on only first-hand information, introducing secondary response, and stressing the importance of identity.