Efficient decentralized consensus protocols in a distributed computing system

Two classes of efficient decentralized consensus protocols for a distributed computing system consisting of an arbitrary number of nodes, one without an initiator and the other with an initiator, are described. It is shown that the protocol without an initiator can be systematically executed and completed in the minimal number of steps. The protocol with an initiator is divided into three phases: broadcasting phase, shuffling phase, and confirming phase. It is proved that under the protocol with initiator, a distributed system of p nodes reaches consensus with an initiator in the minimal number of steps required. The total number of messages required by the protocol with initiator is derived.<<ETX>>

[1]  Philip S. Yu,et al.  Optimal all-to-all broadcasting schemes in distributed systems , 1991, [1991] Proceedings of the First International Conference on Parallel and Distributed Information Systems.

[2]  Ming-Syan Chen,et al.  Subcube Allocation and Task Migration in Hypercube Multiprocessors , 1990, IEEE Trans. Computers.

[3]  S.-M. Yuan,et al.  A class of optimal decentralized commit protocols , 1988, [1988] Proceedings. The 8th International Conference on Distributed.

[4]  Hector Garcia-Molina,et al.  How to assign votes in a distributed system , 1985, JACM.

[5]  T. V. Lakshman,et al.  Efficient decentralized consensus protocols , 1986, IEEE Transactions on Software Engineering.

[6]  Hector Garcia-Molina,et al.  Consistency in a partitioned network: a survey , 1985, CSUR.

[7]  Arthur L. Liestman,et al.  A survey of gossiping and broadcasting in communication networks , 1988, Networks.