Visible to the public Detection of Sybil Attack on Tor Resource Distribution

TitleDetection of Sybil Attack on Tor Resource Distribution
Publication TypeConference Paper
Year of Publication2020
AuthorsGe, K., He, Y.
Conference Name2020 IEEE International Conference on Power, Intelligent Computing and Systems (ICPICS)
Keywordsanonymous communication system, Approximation algorithms, Bipartite graph, Bridges, Communication systems, composability, computer network security, computer program, detection, detection method, Distribution strategy, enumeration attacks, graph theory, Integer Linear Program, integer programming, Linear programming, Metrics, minimum malicious user, Network security, pubcrawl, Relays, Resiliency, resource distribution, resource distribution process, Resource management, suspicious malicious users, Sybil attack, sybil attacks, tor anonymous communication system, tor resource distribution, Web sites
AbstractTor anonymous communication system's resource publishing is vulnerable to enumeration attacks. Zhao determines users who requested resources are unavailable as suspicious malicious users, and gradually reduce the scope of suspicious users through several stages to reduce the false positive rate. However, it takes several stages to distinguish users. Although this method successfully detects the malicious user, the malicious user has acquired many resources in the previous stages, which reduce the availability of the anonymous communication system. This paper proposes a detection method based on Integer Linear Program to detect malicious users who perform enumeration attacks on resources in the process of resource distribution. First, we need construct a bipartite graph between the unavailable resources and the users who requested for these resources in the anonymous communication system; next we use Integer Linear Program to find the minimum malicious user set. We simulate the resource distribution process through computer program, we perform an experimental analysis of the method in this paper is carried out. Experimental results show that the accuracy of the method in this paper is above 80%, when the unavailable resources in the system account for no more than 50%. It is about 10% higher than Zhao's method.
Citation Keyge_detection_2020