Goal and predictive coordination in two level hierarchical systems

The multilevel system theory applies two general methods of coordination, which influences the manner of subsystem management—goal and predictive coordinations. Both these general types of coordination perform multiple data transfer between the hierarchical levels and delay the evaluation and implementation of a global optimal solution of a control problem. The paper demonstrates a coordination policy, which decreases the information transfer in the hierarchical system, titled “noniterative” coordination. The last is developed both for goal and predictive coordination strategies. The mathematical foundations of these two noniterative coordination strategies are presented. Comparative analysis is performed to identify peculiarities and drawbacks for the real time management of two level hierarchical systems. Assessment of the computational workload and speed of the coordination, expressed as “flops” numbers is done for the case of nonlinear optimization problems. Both the noniterative coordination strategies benefit the real time operation in the multilevel system by reducing the iterative computations and the data transfer between the hierarchical levels. The predictive coordination has potential in speeding the management process and resource allocation, due to the decomposition approach, which is applied.