Distributed finite-time average-consensus with limited computational and storage capability

Consensus is a fundamental feature of distributed systems, and it is the prerequisite for several complex tasks, such as flocking of mobile robots, localization in wireless-sensor networks, or decentralized control of smart grids. Average consensus, in particular, is quite challenging, because it is typically obtained asymptotically, while few finite-time algorithms are available. In this paper, we provide a methodology to achieve distributed average consensus in finite time, while maintaining low computational and memory requirements, and small completion times. The provided solution, namely, finite-time average-consensus by iterated max-consensus (FAIM) is based on several runs of the max-consensus algorithm, and has low memory requirements for each node. Compared to existing Flooding approaches, the proposed algorithm requires less memory, at the cost of a slight increase in the number of steps required for termination. The FAIM algorithm assumes that the nodes are aware of an upper bound on the network diameter. To relax this assumption, we complement this paper with a novel distributed algorithm that, in the case of undirected graphs, provides an upper bound on the network diameter which, in the worst case, is twice the actual diameter. A comparison of the proposed finite-time algorithm against the state of the art concludes this paper.