Spreading the control complexity in decentralized control of interconnected systems

The problem is considered for large scale interconnected systems of designing decentralized dynamic controllers so that the sum of the orders of the local controllers is approximately the system order, while the separate controller orders are more evenly spread than permitted by the procedure of Corfmat and Morse, which generally leads to all memoriless controllers, save one of high complexity. The paper shows that under certain circumstances, it is possible to design dynamic feedback controllers for each channel which introduce stable uncontrollable or unobservable modes to the overall system, thus lowering the effective order of that system. This result is then used to spread the controller complexity.