Date of Award

3-26-2020

Document Type

Thesis

Degree Name

Master of Science in Computer Science

Department

Department of Electrical and Computer Engineering

First Advisor

Robert C. Leishman, PhD

Abstract

Performing flight tests is a natural part of researching cutting edge sensors and filters for sensor integration. Unfortunately, tests are expensive, and typically take many months of planning. A sensible goal would be to make previously collected data readily available to researchers for future development. The Air Force Institute of Technology (AFIT) has hundreds of data logs potentially available to aid in facilitating further research in the area of navigation. A database would provide a common location where older and newer data sets are available. Such a database must be able to store the sensor data, metadata about the sensors, and affiliated metadata of interest. This thesis proposes a standard approach for sensor and metadata schema and three different design approaches that organize this data in relational databases. Queries proposed by members of the Autonomy and Navigation Technology (ANT) Center at AFIT are the foundation of experiments for testing. These tests fall into two categories, downloaded data, and queries which return a list of missions. Test databases of 100 and 1000 missions are created for the three design approaches to simulate AFIT's present and future volume of data logs. After testing, this thesis recommends one specific approach to the ANT Center as its database solution. In order to enable more complex queries, a Genetic algorithm and Hill Climber algorithm are developed as solutions to queries in the combined Knapsack/Set Covering Problem Domains. These algorithms are tested against the two test databases for the recommended database approach. Each algorithm returned solutions in under two minutes, and may be a valuable tool for researchers when the database becomes operational.

AFIT Designator

AFIT-ENG-MS-20-M-045

DTIC Accession Number

AD1103020

Download

Share

COinS