Time-based Consensus

International audienceReaching consensus is fundamental in distributed computing. For each execution of a consensus algorithm, there is no difference between the proposed values by different nodes with respect to their proposed times. By presenting a realistic application scenario related to distributed asynchronous mobile robots in dynamic environments, we argue some safety-critical, real-time systems require reaching consensus on the newest proposed values when the old proposed values may not be valid anymore. Afterward, we formulate a new type of consensus problem called time-based consensus, which requires to take into account the times of proposed values. Finally, to tackle such a consensus problem, we determine an essential characteristic which should be considered