Follow

• 

  Exploring RNNs for analyzing Zeek HTTP data

  Daniel Andrews, Jennifer Behn, Danielle Jaksha, Jinwon Seo, Madeleine Schneider, James Yoon, Suzanne J. Matthews, Rajeev K. Agrawal, and Alexander Mentis

  Cyber vulnerabilities pose a threat across systems in the Department of Defense. Finding ways to analyze network traffic and detect malicious behavior on a network will help keep these systems safe. This poster looks at the data collection techniques, model creation, and results of building a recurrent neural network to classify incoming traffic as normal or malicious. Additionally, it considers how the information will be best portrayed on a GUI to network administrators. The model's initial accuracy is 83.45% when trained on 500,017 connections. With increased accuracy, this tool may be used by the Department of Defense to help defend its networks.

• 

  The Adventures of ScriptKitty: Teaching middle school students cyber awareness with comics on the Raspberry Pi

  Ovidiu-Gabriel Baciu-Ureche, Carlie Sleeman, Karlee Scott, William C. Moody, and Suzanne J. Matthews

  Cyber security and on-line safety practices are not commonly taught in schools. However, there is an increasing need for education in these topics as children are joining the Internet community at a much earlier age than previous generations. It is crucial that young people understand the risks they may face on-line and how to mitigate them, ideally as soon as they begin using the Internet unsupervised. The Adventures of ScriptKitty (AOSK) introduces students to basic cyber security concepts using the Raspberry Pi, a single board computer that retails for $35.00. We created AOSK to help facilitate a culture of good cyber security practices and raise interest in STEM. The material is presented in the form of comics paired with instructional sections, including sections of more detailed technical information for readers who wish to learn more about key concepts. We piloted a portion of AOSK to a group of local middle school students. Our time with the students was limited, so we administered a short quiz, then discussed the Raspberry Pi. Next, students completed the packet sniffing exercise from Chapter 2, with the authors available to answer questions and help troubleshoot. Students were asked to re-take the quiz afterward. Our preliminary results show that students achieved a greater understanding of the material, with improved scores of 14%. A custom Pi image preloaded with Kali Linux and all needed software is included with the material. All the materials are published and available for free through GitBook at: https://suzannejmatthews.gitbooks.io/aosk/content

• 

  The Army's ICS and its Defense

  Ray Blaine

  The United States Army is a massive organization with an incredible number of diverse networks. These networks range from unclassified and classified business networks to industrial controls system (ICS) networks in critical infrastructure. As the team lead for one of the first cyber protection teams to attempt to tackle ICS security for the DoD I came across significant issues regarding tools, personnel (team composition, talent, retention, etc.), as well as typical organizational inertia. My team was able to overcome many of these challenges, but there are significant obstacles and opportunities for the larger community to tackle. This talk will elaborate on some of the defensive techniques used in ICS networks and how they differ from traditional IT networks, focusing on some of the challenges unique to ICS networks and military operations to secure them. Finally, a few of the United States Military Academy’s educational efforts in this field will be highlighted.

• 

  Designing a Raspberry Pi sensor network for remote observation of wildlife

  Brian H. Curtin, Rachelle H. David, Emmet D. Dunham, Cullen D. Johnson, Nikhil Shyamkumar, Thomas Babbitt, and Suzanne J. Matthews

  Scientists and the military need unobtrusive methods of observing wildlife. In this poster, we assess the feasibility of a Raspberry Pi sensor network for wildlife detection and monitoring. While technology for wildlife observation such as camera traps exist, they are expensive and/or require human intervention to collect the data. An inexpensive sensor network that can take pictures of wildlife with negligible human intervention will enable scientists and military personnel to discretely detect the presence of endangered wildlife. Each raspberry pi sensor node collects data for transmission to a remote user via an android app interface. In order to improve battery efficiency, the system features an adaptable network with four distinct sleep modes. We also explore increasing the range and durability of the network by integrating a mesh network.

• 

  How Robots and Autonomous Weapon Systems are Changing the Norms and Laws of War

  Christopher Korpela, Dominic M. Larkin, David Parsons, and William J. Barry

  An interdisciplinary team of military and civilian professors of philosophy, electrical engineering, computer science, and robot engineers in the Robotic Research Center at the United States Military Academy at West Point are currently teaching a cutting-edge interdisciplinary project designed for cadets to explore, both in the classroom and in robotic lab environments, artificial intelligence (AI) powered robots, drones, self-driving cars, and emerging human on and out of the loop technologies such as robot swarms. Cadets learn how the increasing sophistication and autonomous decision-making capabilities of AI robots and autonomous weapon systems (AWS) on the battlefield, and in conflict operations, is disrupting existing status quo legal and moral norms and requires rethinking the sacrosanct idea of traditional Just War Theory as the moral compass for justice in declaring and fighting a war. The unpredictable pace of change is revolutionizing the concepts of just war, agency, and human purpose in war.

• 

  Maximizing Computing on Minimal Hardware

  Suzanne J. Matthews

  Single board computers (SBCs) such as the Raspberry Pi have gained popularity in recent years. But, the multicore capabilities of SBCs make them powerful platforms for energy-efficient computation. In this talk, I discuss how we can leverage the multiple cores of an SBC to solve certain problems faster (and at lower total energy) than normal computers, and how to use SBCs to engage and inspire the next generation of computer scientists.

• 

  Row Row Row Your Bot – Trust Me or Not: Investigating the Performance of Human-Robot Teams

  Ericka Rovira, Dominic Larkin, Michael Novitzky, Nikiay Comer, Kaley Rose, and Britany Van Lange

  In this work we present the latest results in Project Aquaticus manned-unmanned teaming looking at the interactions of robot autonomy, human-robot trust, robot reliability, and participant executive attention capability. Project Aquaticus enables manned-unmanned teaming research in the marine domain. It creates exciting and stressful environments for participants based on playing games of capture the flag on the water. For these experiments, a participant was teamed up with an autonomous robot teammate. For consistency, they played against a team composed of two autonomous robots. The independent variables included autonomous teammate reliability, autonomous teammate autonomy level, and task load based on opponent tactics. Forty-eight Cadets at the United State military Academy, West Point, NY played Project Aquaticus simulations using our simulation engine and a game controller. Participants performed executive attention tasks and then played four rounds of capture the flag with several questionnaires interspersed including the NASA TLX and Schaefer trust scale. We present the adaptations to the autonomous robot teammate and opponent AI for the reliability and autonomy level of the autonomous robot teammate and the scenarios performed by the opposing team. We also present the general findings of team performance based on reliability, autonomy level, and scenario difficulty.

• 

  Exploring the raspberry Pi for data summarization in wireless sensor networks: poster

  Andres Alejos, Matthew Ball, Connor Eckert, Michael Ma, Hayden Ward, Peter Hanlon, and Suzanne J. Matthews

  Single board computers (SBCs) are a class of devices where the entirety of the computer is printed on a single circuit board. The Raspberry Pi is perhaps the most popular SBC on the market today. The Raspberry Pi version 3 (1.2 Ghz A53 CPU, 2 GB of RAM), and the Raspberry Pi Zero W (1.0 Ghz ARM11 CPU, 512 MB RAM), cost $35.00 and $10.00 respectively, and both include integrated wireless and Bluetooth. Unlike microcontrollers, SBCs are fully functioning computers with more memory and processing power than the typical sensor. Their powerful System-on-a-Chip (SoC) processors make SBCs good candidates for at-node data summarization tasks in a wireless sensor network [1]. Reducing data transfer in a wireless sensor network is critical for energy efficiency and improved latency [2]. In this poster, we explore the viability of a wireless sensor network composed of Raspberry Pis for video and audio summarization tasks. Our contributions include a i.) novel sensor and gateway node design and ii.) a user interface implemented as an Android App.

• 

  Integrating historical and real-time anomaly detection to create a more resilient smart grid architecture: poster

  Spencer Drakontaidis, Michael Stanchi, Gabriel Glazer, Antoine Davis, Madison Stark, Caleb Clay, Jason Hussey, Nicholas Barry, Aaron St. Leger, and Suzanne J. Matthews

  Ensuring the security of the power grid is critical for national interests and necessitates new ways to detect power anomalies and respond to potential failures. In this poster, we describe our efforts to develop and optimize analysis methodologies for a 1000 : 1 scale emulated smart grid at the United States Military Academy [2]. In contrast to previous work [3, 4], we explore historical analysis using Apache Spark [5] and integrate a Raspberry Pi into our testbed for real-time anomaly detection. We also implement a software controlled physical event and fault generator to induce and measure faults. Figure 1 gives an overview of our system.

• 

  Leveraging Single Board Computers for Low-Energy Computation

  Suzanne J. Matthews

  As power becomes a forcing function in computer design, scientists are forced to explore more energy-efficient architectures. In this talk, I make the argument for single board computers as a low-cost, energy-efficient alternative for certain categories of data- and compute-intensive tasks. In particular, I discuss the successful use of single board computers for anomaly detection in the smart grid. Additional applications in education are also discussed.

• 

  Investigating tensorflow for airport facial identification: poster

  Nikolay Shopov, Mingu Jeong, Evin Rude, Brennan Neseralla, Scott Hutchison, Alexander Mentis, and Suzanne J. Matthews

  Facial recognition is a rapidly developing application of machine learning. Face identification is specifically being adopted across security systems such as airports, perimeter security, and law-enforcement. In this poster, we describe a facial identification approach that can be deployed at airports. Our contributions include i.) facial identification software built on top of Google's TensorFlow [1] framework; ii.) a data collection scheme that can be implemented at airports nationally; and iii.) a user interface for collecting data.