ROTC-Workforce Cyber Security Training Program
Abstract
ABSTRACT:Cyber security threats have has grown dramatically over the past few years and are expected to increase exponentially as we go forward. They especially pose a major threat to our national security. Because of these threats, the need for military engineers trained in cyber analysis will be unparalleled. The objective of this project is to provide training in cyber security analysis tools and techniques to our ROTC cadets so as to provide a highly skilled, diverse pool of military engineers to help fill this future void. Over the last few years, machine learning for cybersecurity has experienced rapid growth, rising from an almost non-existent presence to currently being implemented in over half of cybersecurity techniques utilized commercially. We plan to study new machine learningtechniques and examine how they can be more effectively utilized in cybersecurity. Machine learning is a very important concept to introduce to the cadets. The general idea that a less than optimal solution may be better if it can be obtained in a fraction of the time, is a valuable problem solving concept to understand. Since so much of the latest technology is utilizing machine learning and AI, it is essential that the cadets gain experience working with these systems at the development level. Additionally, in this work a greater emphasis was placed on the cadets~ experience with modern software development and computer programming techniques. This will be very useful in their careers as they interact with less commercial, more open source software.We also plan to study how these cybersecurity techniques can be more effectively implemented through specialized hardware to reduce the Size, Weight, and Power (SWaP) of handheld and mobile systems. A very large number of systems used by soldiers are small internet connecteddevices that need to be secured against various cyber-attacks, but cannot run complex machine learning algorithms due to power constraints. Specialized hardware is needed to make them more secure against cyberattacks and yet meet the SWaP constraints for mobile use. When comparing our plan of work to that of previous years, an increased focus was placed on researching methods for moving our machine learning cyber security systems to low powerembedded processors. We also plan to study a wider selection of neural network algorithms to ensure we are utilizing the best software possible to accomplish this task. Furthermore, we intend to utilize sensors and sensor processing to propose alternative novel methods for increasing security through user authorization and face recognition. Lastly, we plan to use our substantialbackground in computational methods for decision making to analyze the combinatorial aspects of the problems in cybersecurity to determine the effectiveness of an autonomous cognitive agent for use in network intrusion detection.The impact of this research project is that it will create a toolkit of proven functional machine learning algorithms for network security systems. These algorithms can be implemented on large data centers or on mobile systems using neuromorphic hardware for SWaP improvements. This toolkit can then be used to study complex problems in cybersecurity and datamining. Severaldifferent learning algorithms can be evaluated under different conditions to determine the best possible options for cyberattack detection and data mining.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 16, 2019
- Source ID
- N000142012028
Entities
People
- Charles Browning
Organizations
- Office of Naval Research
- United States Navy
- University of Dayton