Security situation assessment in UAV swarm networks using TransReSE: A Transformer-ResNeXt-SE based approach

Abstract

With the rapid development and extensive application of unmanned aerial vehicles (UAVs), the issue of UAV swarm network security has become prominent. To protect the security of UAV swarm networks, effective network security defense measures are crucial. One key aspect is the assessment and monitoring of the network's security situation. However, most existing research focuses on the security of individual UAVs or detecting specific attacks, which fails to provide proactive protection for the network. To address this issue, we propose a UAV swarm network security situation assessment method, which combines the Transformer network with the optimization of the Aggregated Residual Transformations for Deep Neural Networks (ResNeXt) and squeeze-and-excitation (SE) structure (named TransReSE). By using multiple scale-cross convolution kernels, TransReSE can efficiently extract data features and improve situation assessment accuracy through the Transformer network. Experimental results from four public datasets have shown that TransReSE outperforms other schemes in terms of accuracy, recall, and F1. By assessing the value of the swarm network situation and the threat level, we can make faster, more effective decisions and proactively allocate resources to defend against UAV swarm network attacks.