Reducing communication overhead for average consensus

An average consensus protocol is an iterative distributed algorithm to calculate the average of local values stored at the nodes of a network. Each node maintains a local estimate of the average and, at every iteration, it sends its estimate to all its neighbors and then updates the estimate by performing a weighted average of the estimates received. The average consensus protocol is guaranteed to converge only asymptotically and implementing a termination algorithm is challenging when nodes are not aware of some global information (e.g. the diameter of the network or the total number of nodes). In this paper, we are interested in decreasing the rate of the messages sent in the network as nodes estimates become closer to the average. We propose a totally distributed algorithm for average consensus where nodes send more messages when they have large differences in their estimates, and reduce their message sending rate when the consensus is almost reached. The convergence of the system is guaranteed to be within a predefined margin η. Tuning the parameter ! provides a trade-off between the precision of consensus and communication overhead of the protocol. The proposed algorithm is robust against nodes changing their initial values and can also be applied in dynamic networks with faulty links.