Synchronization of autonomous agents by an optimal networked controller

This paper investigates the synchronization of identical linear agents based on optimal control methods. Besides asymptotic synchronization the transient behavior of the networked multi-agent system is required to minimize an LQ-like performance index which is chosen so as to reflect the motion of the synchronization errors. The resulting optimization problem is non-convex because the networked controller has to satisfy structural constraints. However, a necessary condition for optimality is derived and it is shown that determining the optimal networked controller is computationally expensive. Therefore, an algorithm is proposed that uses a gradient method to solve the optimization problem for a given initial networked controller.