New framework of back diffusion-based autonomous decentralized control and its application to clustering scheme

We previously proposed a framework for autonomous decentralized control by introducing a partial differential equation based on local-action theory. Although each node in the network acts autonomously on the basis of only the local information directly available to it, our control proposal leads to optimal performance for the whole network. In addition, we have proposed flow control based on a diffusion equation to achieve autonomous load-balancing in the network and have confirmed that realizes congestion avoidance. This paper introduces a new autonomous decentralized control scheme that creates the spatial structures of finite size by using renormalization transformation and back diffusion drift. Moreover, we use the proposed control scheme to realize autonomous decentralized clustering in ad hoc networks.