Distributed average consensus: Beyond the realm of linearity

In this paper, we present a distributed average-consensus algorithm with non-linear updates. In particular, we use a weighted combination of the sine of the state differences among the nodes as a consensus update instead of the conventional linear update that just includes a weighted combination of the state differences. We show the non-linear average-consensus converges to the initial average under appropriate conditions on the weights. By simulations, we show that the convergence rate of our algorithm outperforms the conventional linear case.