Global Sensor Management: Allocation of Military Surveillance Assets

Show full item record

Title: Global Sensor Management: Allocation of Military Surveillance Assets
Author: Arney, Kristin M
Advisors: Thom J. Hodgson, Committee Chair
Russell E. King, Committee Member
E. Downey Brill, Committee Member
Tim Trainor, Committee Member
Abstract: The United States uses an integrated missile defense system this system to detect, track, classify and destroy any threats from the air. This system can be represented by a network of nodes and arcs where the nodes are depictive of the sensors and their required functions and the arcs detail the ability of a sensor to complete a function and pass along the appropriate information to a machine in a subsequent stage in the operation. Given a set of sensor resources, we need to develop a management tool that assigns resources to tasks and functions to maximize both our ability to detect and eliminate threats and our sensor coverage of various regions of earth and space, and that tool needs to be flexible enough to deal with continuous changes to the environment in which it is operating. To that end, we investigate and develop both a total enumeration method generating all possibilities and determining the optimal using our own developed methodology implemented through a Visual Basic code. This brute force method in order to find optimal assignments is not the preferred method due to the potential time constraints associated with scaling up the network and developing a real time decision. However, in this research, we focus on an operational planning model which assists us in understanding the complexities and sensitivities of the problem. Throughout the project, STRATCOM's 4-task sample network is used as a test case. Probability values are randomly generated to be representative of true classified data. In solving the sample network for the static case for missile defense type event, we use a model calculating probabilities based on the Law of Total Probability. Using total enumeration, we are to run all possible allocations for the assignable assets and determine all associated endstage probabilities. From here, we can assign the assets to the appropriate task according to any constraints provided. Constraints could be both lower bounds for tasks 2, 3, and 4 as well as specifications for the endstage probabilities for machines in stage 10 of task 1. Results suggest that the sample network is both easy to solve and very insensitive. We assume that models to solve this instance can be applied to higher order problems with additional assignable assets, different network connections, as well as multiple events.
Date: 2008-05-08
Degree: MS
Discipline: Operations Research

Files in this item

Files Size Format View
etd.pdf 1.691Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record