eSHANCS: Evolving Simulated Human Activity on Networked Computer Systems
Department
Computer Science
Major
Computer Science
Research Advisor
Tauritz, Daniel R.
McMillin, Bruce M.
Advisor's Department
Computer Science
Second Advisor's Department
Computer Science
Funding Source
National Science Foundation’s Scholarship for Service
Abstract
Emulated computer networks are commonly used in cyber security research to evaluate attacker and defender strategies, but it is often trivial to detect adversaries in the absence of complex human user interactions, because irregular activity can be attributed to an adversary. The research proposed is to create a custom hyper-heuristic employing genetic programming that evolves computer agents capable of simulating the complex behavioral patterns exhibited by human users on a diverse set of networks. Previous research has been mostly limited to replays of user traffic which does not generalize well to different networks. The proposed approach will use anonymized, enterprise scale network traffic data sets provided by Los Alamos National Laboratory to train and test the system. The goal will be to have the entire population of agents act as a multi-agent simulation of human users with sufficient fidelity to provide network emulations capable of representing the real world.
Biography
Joshua Herman is an undergraduate student expecting to graduate with his B.S. in Computer Science at Missouri University of Science and Technology in May 2016. After graduating he will pursue a Ph.D. in Computer Science with an emphasis on cyber security and evolutionary computing at Missouri University of Science and Technology through the National Science Foundation’s Scholarship for Service. He is currently an undergraduate research assistant in the Natural Computation Laboratory supervised by Dr. Daniel Tauritz and Dr. Bruce McMillin. He has had two summer internships along with working year-round through telecommuting at Sandia National Laboratories.
Research Category
Research Proposals
Presentation Type
Poster Presentation
Document Type
Poster
Location
Upper Atrium/Hallway
Presentation Date
11 Apr 2016, 1:00 pm - 3:00 pm
eSHANCS: Evolving Simulated Human Activity on Networked Computer Systems
Upper Atrium/Hallway
Emulated computer networks are commonly used in cyber security research to evaluate attacker and defender strategies, but it is often trivial to detect adversaries in the absence of complex human user interactions, because irregular activity can be attributed to an adversary. The research proposed is to create a custom hyper-heuristic employing genetic programming that evolves computer agents capable of simulating the complex behavioral patterns exhibited by human users on a diverse set of networks. Previous research has been mostly limited to replays of user traffic which does not generalize well to different networks. The proposed approach will use anonymized, enterprise scale network traffic data sets provided by Los Alamos National Laboratory to train and test the system. The goal will be to have the entire population of agents act as a multi-agent simulation of human users with sufficient fidelity to provide network emulations capable of representing the real world.
