Finite-time distributed averaging

This paper proposes a distributed averaging algorithm for multi-agent networks, in which each agent is with a real-valued measurement. Provided that the underlying graph of the network is a tree, the proposed algorithm enables each agent to compute the average of the values of all agents in the network in a finite number of steps. Different from most existing finite-time distributed averaging algorithms, the algorithm proposed in this paper does not require each agent to know any global information.