notesum.ai

Published at October 21

Designing Robust Cyber-Defense Agents with Evolving Behavior Trees

cs.CL

cs.AI

cs.LG

Released Date: October 21, 2024

Authors: Nicholas Potteiger¹, Ankita Samaddar¹, Hunter Bergstrom¹, Xenofon Koutsoukos¹

Aff.: ¹Department of Computer Science, Vanderbilt University

Arxiv: https://arxiv.org/abs/2410.16383v1

Refer to caption

Notation	Description
$v_{f}\in V$	fire source
$T_{burn}^{max}\in\mathbb{N}$	maximum burning time of any node
$T_{burn}^{t}(v)\in\mathbb{N}$	burning time of node $v$ at timestep $t$
$T$	total number of timesteps upto which the game is played
$r\in\mathbb{N}$	drone search radius
$R_{t}\subset V$	set of retardant nodesa at timestep $t$ , i.e., set of nodes that contain the “fire” spread using flame retardant
$I_{t}\subset V$	set of visibility nodes at timestep $t$ , i.e., set of nodes that are visible from the deployed drones
$D_{t}(v)\in\{-1,0,1\}$	status of a drone deployed on node $v$ at timestep $t$ , ( $-1:$ no drone deployed on node $v$ , $0:$ drone is deployed but not activated, $1:$ drone action is activated)