Internet of Battlefield Things: Challenges, Opportunities, and Emerging Directions

Dynamic reinforcement learning for network defense: botnet detection and eradication

“…What differentiates the DoD space from the commercial space at large is the existence of computer networks safeguarding national security information and computer systems whose untimely failure could have especially Send correspondence to: rschabinger@dci-solutions.com catastrophic consequences on the modern battlefield. 4 It follows that both DoD enterprise networks and military combat-oriented systems in the field make attractive targets for APTs; compromised enterprise networks would leave them wide open to the theft of sensitive data, whereas compromised combat systems could potentially be rendered inoperable at crucial moments on the battlefield resulting in the loss of life. In this paper, we initiate the study of this class of threats by considering an APT whose goal is to first gain a foothold on a DoD enterprise network through social engineering and then perform lateral movements to establish a clandestine botnet, which ultimately spans the entire network.…”

Autonomous cyber warfare agents: dynamic reinforcement learning for defensive cyber operations

“…Although this lack of flexibility impacts many organizations, it could have devastating effects on the military's ability to achieve information dominance on the modern battlefield. As the United States Army pivots to more interconnected devices and platforms at the edge of its networks, 4 cyber defenders need the ability to sense, understand, decide, act, and assess more quickly than their adversaries to counter threats. 5 This becomes increasingly difficult as computer networks increase in size and scope while resources remain pooled in enterprise CSOCs.…”