GAUR: a method to detect Sybil groups in peer-to-peer overlays

Abstract

Sybil attack is an important problem as the peer-to-peer networks grow in size and become prominent means for distributing multimedia. In order to validate the feasibility of using psychometric tests as an approach to detect Sybils in their entirety as Sybil groups, an experiment is conducted by taking the tests on a population of considerable size. Selected people were given multiple questionnaires corresponding to multiple identities in the network. The survey data is analysed using DBSCAN clustering algorithm using several metrics. The results show that 75% of the Sybil groups were detected with 67% completeness.