Journal of Dali University ›› 2023, Vol. 8 ›› Issue (6): 24-32.

Previous Articles     Next Articles

MV-Raft A New Consensus Algorithm forAllianceChain Based on Multidimensional Vectors

Wu Guifeng Hu FaZeng Xin Yang Dengqi Li Xiaowei*   

  1. College of Mathematics and ComputerDali University Dali Yunnan 671003China
  • Received:2022-10-18 Revised:2022-11-16 Online:2023-06-15 Published:2023-06-26

Abstract:

The default algorithm of Raft for existing alliance chains uses a random mechanism for node crashes without considering that different node weights may result in longer consensus times and greater log synchronization pressures on leader nodes. To address this issue a Raft algorithm based on multidimensional vectors is proposed called MV-Raft. The algorithm first constructs an eigenvalue vector according to factors that affect the efficiency of reaching a consensus and then broadcasts the above eigenvalue vector to the whole network when the distributed system is started or a single node is newly added to the system. The similarity between nodes is calculated based on the eigenvalue vector and the n nodes with the highest similarity are cached within each node for prioritized operation during the election and synchronization process. The simulation results show that MV-Raft algorithm can reduce the time consumption of the whole process by 16% and reduce the pressure of a single leader node by 73% at most.

Key words:

"> blockchain">, alliance chain">, consensus mechanism">, Raft algorithm">, multidimensional vector algorithm

CLC Number: