Abstract
This paper presents a theoretical and numerical analysis of the stability of a decentralized swarm of unmanned aerial vehicles under multivector network attacks (DDoS and MITM) and communication delays. The dynamics of each agent are described by a three-dimensional double-integrator model with constraints on velocity and acceleration. The primary control objective is the strict enforcement of the safety constraint ๐๐ ๐๐ โฅ ๐๐ ๐๐๐, even at the expense of formation-keeping accuracy. To ensure the invariance of the safe set, the method of Exponential Control Barrier Functions (ECBF) is employed, implemented in the form of quadratic programming (QP filtering), as well as a variant with slack variables ensuring feasibility under constraint conflicts. The critical delay is evaluated based on the analysis of a linearized time-delay system and compared with an empirical safety threshold. Numerical experiments for two swarms of three agents each (six in total) demonstrate that the ECBF layer maintains safety at delays on the order of 1 s even under DDoS+MITM attacks, while ฮด(t) serves as an indicator of approaching the physical infeasibility of safety.