Self-Securing Systems

Abstract

The pervasive nature of software-reliant systems in today’s modern military creates new opportunities for sophisticated adversaries. The vast majority of DoD weapons systems, platforms, and networks rely on software to operate. Software can often be disrupted remotely, which necessitates a new kind of security to protect against such cyber-attacks. Defending the software and network-based aspects of critical weapon systems is challenging for a number of reasons, chief among which is the advanced nature of the adversary in the cyber realm. Future cyber adversaries will be well-funded, well-informed, and agile. Building weapon systems, platforms, and networks that defend themselves in real time will be vital in protecting ourselves against this adversary. The Department needs systems that can autonomously monitor and manage their own health and security posture through advanced sensing and perception, reasoning, and planning. Such systems identify and classify threats much more quickly than a human operator, and therefore, neutralize the threat more quickly and effectively. However, researchers must be cognizant of the potential unintended consequences of turning security over to autonomous systems. Verification techniques must be developed to ensure that autonomous and dynamic system changes maintain correct mission-focused capabilities without introducing unintended vulnerabilities. Conversely, developing techniques to track and audit actions taken by autonomous systems ensures that direct control can be reasserted.

Document Details

Document Type
Accomplishment
Publication Date
Oct 01, 2020
Source ID
eceb9ccedb5c90f4518e6036c0cedffb

Tags

Fields of Study

  • Computer science

Readers

  • Economics
  • Irregular Warfare and Special Operations Cyberspace Operations against Adversarial Threats.
  • Software Engineering.

Technology Areas

  • Autonomy
  • Autonomy - UAVs
  • Cyber

Related Documents