-
Views
-
Cite
Cite
Amir Jalaly Bidgoly, Behrouz Tork Ladani, Modelling and Quantitative Verification of Reputation Systems Against Malicious Attackers, The Computer Journal, Volume 58, Issue 10, October 2015, Pages 2567–2582, https://doi.org/10.1093/comjnl/bxu130
Close - Share Icon Share
Abstract
In recent years, trust and reputation systems have gained much interest in many environments, such as e-market-places, web services, ad-hoc networks and multi-agent systems. Reputation models are responsible for computing the reputation rank of entities in a community or network based on collecting the opinions. Despite the popularity of reputation systems, they are vulnerable to different kinds of attacks. These attacks, which are a sequence of misleading behaviour performed by malicious entities, can simply lead the system to erroneous results. In this paper, we propose a novel method for quantitative verification of reputation systems against these types of attacks. In the proposed method, a reputation system is formally defined using three related models: reputation model, honest entities model and attacker model. The attackers are assumed to be agents who want to maximize their received rewards by abusing the system. The system is then formally evaluated using Markov Decision Process framework. The proposed method is capable of verifying the reputation systems against predefined attacks as well as discovering unknown attacks. The method can also find the worst possible attack plan against a given system. To illustrate the applicability of the proposed method, two case studies are presented for analysis and comparison of Beta and eBay reputation models.
