Optimal decentralized control

The main objective of this paper is to present a solution of the longstanding problem of optimal decentralized control using the classical method of Lagrange. The key idea of the paper is to formulate decentralized information structure constraints as differential equations, which are added to the equations of motion to form a suitable set of constraints for minimization of a cost functional. The globally optimal solution is obtained by introducing Lagrange-Lyapunov functions as multipliers and applying Pontryagin´s maximum principle. By assuming a Gaussian nature of the state evolution we provide a feedback control structure determined by Riccati-type equations, which is the basic feature of the classical result of Kalman, and is of major practical significance.