White Papers
Discrete System Optimization Using Cost As Independent Variable (CAIV) Techniques With Distributed Analysis Models
Brett Malone and Scott Woyak
Abstract
A design environment has been constructed to assist designers and analysts in performing concept studies and optimization based on distributed analysis to support Cost As Independent Variable (CAIV) studies. This environment uses application server-based techniques to host distributed simulations and analysis tools. These components are then linked through a common client interface to build an integrated model. A set of unique design exploration tools provides the analyst with a decision support environment for performance, cost, and operational assessments. A design example based on this architecture is shown. This example illustrates discrete configuration selection of platform and sensor types based on a non-gradient based optimization algorithm to defeat various threats given system cost constraints.
Introduction
Network Centric Warfare represents a desired capability by military agencies to combat emerging threats around the world. This new doctrine introduces complexities because systems must be evaluated based on traditional physical disciplines for performance as well as operational concepts such as node-to-node connectivity and subsystem cost and interoperability. Cost as Independent Variable (CAIV) tools used to design network centric systems must support the conceptual phase of a program by providing analysts with a trade space evaluation tool for these complex interactions. Information that is crucial to all systems and subsystems must be linked from distributed models used by a team of designers if these interactions are to be assessed properly. In constructing a design environment to evaluate candidate concepts for network centric systems, the impact of cost, system performance and variability of threats must be considered. The system designer must have the ability to add subsystems, or "nodes," to the battlefield network, increasing the complexity of the problem and requiring more sophisticated search algorithms to analyze the thousands of combinations of each subsystem. In this research, our goal is to give the system designer a capability that clearly selects from discrete choices of components to form an overall system that fits a desired cost constraint and satisfies a minimum performance on the battlefield.
Download The Complete Paper
(A one-time registration will be requested)
