Toward the resilience of UAV swarms with percolation theory under attacks

Abstract

Unmanned aerial swarms have been widely applied across various domains. The security of swarms against attacks has been of significance. However, there still exists a lack of quantitatively assessing the unmanned swarm resilience against attacks. Thus, this work adopts the percolation theory to mathematically analyse the resilience of the unmanned aerial swarms after random attacks. In addition to the typically used popularity in the preferential attachment, distance of neighbours is taken into account for modelling unmanned swarms, which is missing in the literature. This improved preferential attachment-based swarm model offers a more precise and realistic description of swarm behaviours. In addition, an attack model is proposed, which can be a description of dynamic attacks. Moreover, this study also utilizes the percolation theory to assess the resilience of swarms after the random attacks. Finally, the simulation results show that the resilience derived using percolation theory aligns with the improved swarm model. The proposed swarm model maintains $79\text{\%}$ resilience when $20\text{\%}$ of the UAVs are attacked under random attacks, and even $69.4\text{\%}$ resilience when $20\text{\%}$ of the UAVs are attacked under initial betweenness-based attacks.