Improved Raft Consensus Algorithm in High Real-Time and Highly Adversarial Environment

Abstract

High real-time and highly adversarial environment put forward higher requirements for the performance of blockchain consensus algorithm. To improve Raft’s consensus efficiency and safety, we propose an improved Raft algorithm called “hhRaft” to optimize Raft consensus process by introducing a new role of monitor. In the leader election phase, monitor nodes supervise the candidate nodes by identifying the malicious node’s forged Requestvote message. In the log replication phase, monitor nodes supervise the leader node by comparing the computing results of transactions. Through the performance test on the Consortium Blockchain -- Hyperledger Fabric, it is proved that hhRaft is superior to the original Raft algorithm in terms of transaction throughput, consensus latency, and anti-Byzantine Fault capabilities, making it suitable for use in high real-time and highly adversarial environment.