AI and Game Theory based Autonomous UAV Swarm for Cybersecurity

Abstract

Uninterrupted communication is crucial for modern electromagnetic (EM) spectrum operations where successes of situational awareness, defensive and offensive missions necessitate continuous reliance on wireless transmission. Preventing an adversary from dominating cyberspace becomes challenging as rapid technological developments allow state and non-state actors to engage in a broad range of destructive cyber electromagnetic activities (CEMA). Digital threats to communication networks can range from eavesdropping and impersonation attempts to various forms of denial-of-service attacks. In this paper, we present bio-inspired and game theory based flight control algorithms for a swarm of autonomous UAVs. Each UAV considers MANET connectivity, overshadowed ground area coverage and signal strength from interfering mobile radio emitters. Our algorithms use 3D Voronoi tessellations and linear interpolation for EM mapping of local neighborhood as part of decision making process. Simulation experiments in OPNET show that autonomous UAVS require only limited near neighbor communications to maintain a high area coverage overshadowed by the swarm with uninterrupted MANET connectivity. By providing a lightweight solution for rapidly deployable swarm of autonomous UAVS, our flight control algorithms are good candidates for deployment in complex environments in presence of adaptive and mobile sources of EM interference.