From a control theory perspective, ballistic missile defense requires a distributed dynamic non-linear control system. The conventional approach to coordinating the allocation and use of components that are geographically dispersed and functionally diverse is through a central management function that contains missile defense system and element control law. This central manager has to deal with the complexity of each of the cooperating missile defense elements, their operational characteristics and control methods. A centralized system controller design adversely affects reliability, survivability, and maintainability. As an alternative, a collection of less complex neural controllers, each of which controls a mutually exclusive subset of the missile defense elements and is cognizant of overall system need, will be investigated. The Collection Of Homogeneous cOoperating contRollers Toolkit (COHORT) will support the creation of intelligent, adaptive, and cooperating controllers. Distribution of the system-level management and control design has the potential to reduce complexity and improve system reliability and survivability. Phase I will derive the distributed neural controller framework and define the tool set required to support its implementation. Development of the tool set and its application to a ballistic missile defense scenario will occur in Phase II.
Keywords: Distributed Control, Neural Networksadaptive Control, Pid Controllersneural-Based Controldynamic Non
