M.A. Thesis:

Sybil Attack Detection in E-Commerce Network Based on Clustering and Trust Management

Abstract E-commerce is a giant industry which numerous corporations have put their business procedures based on it due to its benefits. But mutual trust among seller and buyer and specially seller reputation is considered as a vital issue in the area of e-commerce. P2P E-commerce programs are very vulnerable against active and passive attacks. These attacks have taken the potential individuals and commercial corporations out competition field with the aim to acquire maximum benefit and minimum loss in e-commerce. When a transaction occurs, these attacks takes place at interactions among commercial peers. The Sybil attack is one of most important attacks in e-commerce environment where peers could have false and forged multiple identities. Most available research works which focus on the trusted certificates and social networks, aren’t able to prevent the Sybil attacks peers from performing transactions. So in the thesis, we suggest a solution for recognizing and overcoming Sybil attacks. The suggested solution detects and overcomes Sybil attacks using clustering mechanism and similarity trust. Our suggested method have been compared to Sybil Trust, Eigen Trust, Eigen Group Trust mechanisms. Finally, the analysis of performance and security indicates that Sybil attacks could be minimized through our suggested similarity trust. :The first test was performed in order to calculate the recognition percentage of malicious nodes which we saw higher performance of our suggested method than other methods. In this test, we increased the number of Sybil peers from 10 to 40 percent and obtained the recognition rate of Sybil attackers through simulation. Based on this test, as the number of Sybil peers increases, the false alarm rate will increase and the rate of Sybil attacker detection will decrease. Then second test has been performed to determine the optimal value of thd, So that the best detection rate occurs when thd =0.6 which 90% attackers have been detected. Because as thd value is greater than 0.6, then the false alarm rate is increasing. Therefore, we select thd value such that will reduce the rigidity and if thd value is lower than 0.6 , high numbers of attacker nodes will not be detected therefore we determine thd value such that more logical calculations will be achieved. :Keywords Sybil attack, e-commerce, trust, P2P networks.

Keywords: Keywords: Sybil attack, e-commerce, trust, peer-to-peer networks