Scalable and Distributed Control Laws for Network flow Optimization

We investigate the design of scalable and distributed control laws for flow control in a large network – equivalently an optimization problem. We identify the implications of the desired “plug-and-play’ property of of such protocols and propose design principles upon which our algorithm should be built: the class of our algorithms must achieve little extra communication cost and global asymptotic stability. General structural properties of such algorithms are presented and we finally provide an algorithm which satisfies our design principles for a network with heterogeneous delays.