Decentralized adaptive scheduling using consensus variables

In this paper we present it new approach to decentralized adaptive scheduling, using recent results on the control of consensus variables in graphs with nearest-neighbour communication topologies. First, existing results for single consensus variables are extended to include the cases of forced consensus, when one of the negotiating agents is driven by a setpoint, and of constrained consensus, where multiple consensus variables are required to be separated by a fixed amount. Next, we consider a class of adaptive scheduling problems, whereby a set of decentralized coordinators should cooperate to adapt shared schedule times in response to disturbances or changes in the system. Our approach is to choose task timings to be the consensus variables in the system. The utility of these ideas is illustrated using the example of a synchronized strike mission. Copyright (C) 2006 John Wiley & Sons, Ltd.