MCRaft: synergistic collaboration of multi leaders for IoT cluster stability optimization.

Raft is a consensus algorithm to coordinate the proper execution of protocols and rules within the system. Owing to its security and simplicity, it can be applied to the field of IoT with blockchains. When IoT device clusters run under the Raft consensus algorithm, as the number of devices increases, split voting is prone to arise, therefore the Raft algorithm cannot run efficiently under large-scale nodes. To improve the problem of split voting and delay in system recovery after a leader collapse (DISR), this paper proposes a high-fault-tolerant consensus algorithm based on synergistic collaboration of multi leaders within the IoT clusters—MCRaft. The concept of super node cluster is introduced. MCRaft elects three leaders to form a super node cluster mapping single Raft leader. The election of the MCRaft node is verified through the log hash stored by the node, and the voting overhead of the cluster is born by the node’s own computing power instead of relying on the voting mechanism, making the election of leaders more efficient and secure. According to the experiments, the synergistic collaboration of the super node clusters effectively prevents DISR issue hence achieve efficient transmission, improving the fault tolerance of the IoT cluster. MCRaft algorithm improves the election efficiency by about 84% compared with that of common Raft algorithm within 100 nodes. 200 crash experiments indicates that the MCRaft algorithm improves its stability by about 24% on average compared with that of common Raft.