Themis: An Equal, Unpredictable, and Scalable Consensus for Consortium Blockchain

Abstract

Consensus algorithm is the core component of consortium blockchains. Equality, Unpredictability and Scalability are three important demands for the consensus algorithms of consortium blockchain. Existing deterministic consensus algorithms (e.g. PBFT) can ensure Equality, but cannot meanwhile meet Unpredictability and Scalability; probabilistic consensus algorithms (e.g. PoW) can achieve Scalability and guarantee a decent Unpredictability, but cannot meet the Equality requirement. In this paper, we propose a new consensus algorithm, namely Themis, which takes the three properties into account. Themis independently adjusts the block-producing difficulty of each node through a self-adaptive node election mechanism, effectively reducing the correlation between the block-producing frequency and the invested computing power of each node. Besides, a GEOST main chain consensus rule is proposed to handle forks and further improve the performance of the algorithm. If a fork occurs, consensus nodes will choose the sub-chain with the highest Equality to join the main chain. Evaluations show that Themis achieves outstanding performance in Equality and Unpredictability while ensuring Scalability, compared with the existing algorithms.