A Security Detection Model for Selfish Mining Attack

Abstract

As a new technology, the “decentralization” of blockchain has been paid more and more attention. With the increasingly wide range of applications of blockchain technology, its security has become a bottleneck restricting the development. In order to solve the problem, the blockchain structure characteristic can be used to build a security detection model. Taking 51% attack detection as an example, the relationship among attack state, attack intensity and attack time was analyzed. However, because of the variety of blockchain attacks, the test of only one attack cannot verify the effectiveness of the detection model. On the basis of the existing work, this paper studies the security of selfish mining attack under the existing detection model and compares it with 51% attack, and finds that the attack block and honest block are related to the ability of the attacker. When the attacker’s ability exceeds half of the total ability, the attack block is more than the honest block. However, for the number of system states, there is a big difference between 51% attack and selfish mining attack. Specifically, in selfish mining attack, when the attacker’s ability exceeds 25%, the number of states tends to stabilize. While in 51% attack, the attacker’s ability needs to exceed 50% in order to stabilize. This shows that the detection model can well distinguish 51% attack and selfish mining attack. According to this characteristic, we can further construct an effective alarm mechanism.