Detection of Sybil Users Using Random Walk Theory
Large-scale decentralized systems without trusted identities and other peer-to-peer systems are more likely
exposed to Sybil attacks from remote faulty elements that compromise the system’s running with fake information. In such a
Sybil attack, a malicious user pollutes the system by creating multiple fake identities, pretending to be multiple, distinct nodes
in the system called Sybil nodes. Many security mechanisms against such attacks proposed earlier are based on specific
assumptions. In this paper, a novel defense mechanism against the sybil attacks on social network is proposed using random
walk beginning from a known honest node traversing throughout the social graph of the network. Besides we propose social
degree and popularity analysis on the detected honest users obtained from the random walk approach which will eventually
give the list of Sybil users and an action can be taken to block such Sybil users. This paper aims for improved Sybil detection
in the social network as it mitigates attacks from friends of friends, thereby reducing the redundancy of data and fake identities
in the network and also outperforms the running of the network.
Index Terms— Sybil Attacks, Random Walk, Social Degree, Popularity Analysis.