Near-Optimal Design of Reliable Decentralized Stabilizing Controllers for Large-Scale Systems

In the design of efficient schemes for signal processing and decision making in large-scale systems, reliable operation is a factor of major concern since a frequent source of failure is the loss of information transfer between different parts of the system. This paper presents a new iterative procedure for the design of reliable decentralized stabilizing controllers for a large-scale system which can be described as an interconnection of several subsystems. By an efficient exploitation of the trade-off that exists between optimality and reliability, the design procedure yields a decentralized control scheme which not only ensures a high degree of reliability under arbitrary information transfer disruptions but also provides a near-optimal performance. An integration of two distinct approaches one of which is based on an interconnection pattern classification and the other based on a high gain methodology of constructively increaseing the redundancy, are utilized for the development of the step-by-step design algorithm.