Applications of Neural Networks to Command Centers

Continuing increases in computing capability present numerous opportunities to assist the staff of a command center as they analyze enemy operations, construct and evaluate courses of action, and monitor the execution of the selected option. Nowhere are these opportunities more critical than in the areas of national and theatre missile defense. ALPHATECH has developed the most advanced and mature fixed-algorithm decision aid for missile defense, the Look-Ahead Battle Planner, which evaluates the expected outcome of an engagement given certain data about threat and friendly forces. It is based on complex optimization algorithms which schedule interceptor missions against the expected target set. Recent advances in neural computing techniques offer the potential to augment this capability with faster, adaptive techniques which can be rapidly retrained as threat characteristics change. This proposal describes a program to implement and test a neural network which may provide order-of-magnitude increases in responsiveness. Our approach is to generate training and test data sets using the Look-Ahead Battle Planner, to develop and implement a neural network to assess potential battle outcomes given coarse information about threat and friendly forces, and to compare the performance and computational requirements of the neural approach with the existing aid.