Binding efficiency and robustness for blockchains using reputation-based byzantine fault-tolerant consensus algorithms
Gengrui Zhang · 2022
Leader-based Byzantine fault-tolerant consensus algorithms suffer from vulnerabilities during routine view changes as they often follow a predefined schedule to passively rotate leadership among servers. However, the rotation becomes problematic under failures. In particular, when a scheduled server has crashed, the rotation still blindly assigns leadership to that server, causing performance degradation. This problem is exacerbated with further pronounced performance degradation when leadership rotation becomes more frequent. To address this problem, we propose a reputation-bounded active view-change protocol. Each server maintains a reputation state, calculated based on its behavior history, that reflects the likelihood of its correctness. During view changes, servers with a higher reputation are more likely to be elected, suppressing Byzantine faults while increasing the presence of correct leaders.