Enhancing automated red teaming with evolvable simulation

  • Authors:
  • YongLiang Xu;Malcolm Yoke Hean Low;Chwee Seng Choo

  • Affiliations:
  • Nanyang Technological University, Singapore, Singapore;Nanyang Technological University, Singapore, Singapore;DSO National Laboratories, Singapore, Singapore

  • Venue:
  • Proceedings of the first ACM/SIGEVO Summit on Genetic and Evolutionary Computation
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Automated Red Teaming (ART), an automated process for Manual Red Teaming, is a technique frequently utilised by the Military Operational Analysis (OA) community to uncover vulnerabilities in operational tactics. Currently, individual ART studies are limited to the parameter tuning of a simulation model with a fixed structure. The effects in the evolutions of structural features of a simulation model have not been investigated in any of the studies. This paper investigates the benefits of Evolvable Simulation, which involves evolution of the structure of a simulation model. The case study used for this purpose is a maritime based scenario which involves the defense of an anchorage. Simulation results obtained through Evolvable Simulation revealed that the quality of the solutions found given an appropriate amount of evaluations will improve when the simulation model is evolved. Additionally, experimental results also showed that it is likely to have negligible improvement in solutions for models with smaller search space when the amount of evaluations is more than required. The insights obtained in this work shows that evolvable simulation is an effective methodology which allows decision makers to enhance their understanding on military operational tactics.