Unmanned aerial vehicle (UAV) networks are founded upon the low-altitude airspace and are led by the aviation industry. Driven by the fast development of the low-altitude digital economy, new demands and challenges of enhanced security, high collaboration, and regulatability have been raised in UAV networks. Based on the comprehensive review of current research progress on UAV networks, key security threats and challenges faced by UAV networks were explored from four perspectives: behavioral security, communication security, decision security, and data sharing security. A thorough review of the existing and potential solutions was also provided across various aspects, including dynamic anomalous behavior warning, trusted communication link construction, intelligent defense against network attacks, and cross-domain secure data sharing. Finally, several future directions and trends of the UAV networks were outlined about the integration of the UAV networks with the emerging technologies such as semantic communication, large models, and digital twin.

With the booming development of low-altitude economy and unmanned aerial vehicle (UAV) technology, the deep learning based object detection model has been widely used in the field of UAVs, however, with potential security threats in practical deployment. Distinguished from the traditional image classification, the object detection model additionally generates and returns a set of labeled bounding boxes on the basis of identifying objects. Utilizing this feature, a covert backdoor attack framework for object detection model was proposed. Different from the traditional backdoor attack methods that only embed specific trigger features in images, the proposed new framework precisely matched the object categories of images with bounding box labels firstly. Then, based on the matching results, the data and labels were selectively poisoned according to the image scaling principle, achieving high stealthiness of backdoor implantation. Finally, the experimental results demonstrated that the proposed backdoor attack method was with high attack success rate on the real datasets.

With the rapid development of Internet of Things (IoT) technology, the unmanned aerial vehicle (UAV)-assisted edge computing has become crucial for enhancing data processing capabilities and model performance. However, the limited computational and storage capacities of UAV nodes constrain the quality of local models, making them insufficient to support the neural network training tasks effectively. To address this challenge, introducing a federated learning mechanism to construct UAV swarms has proven to be an effective solution. Nevertheless, this approach involves trade-offs between the system availability and the privacy protection, which make poisoning attacks more effective and harder to detect. Traditional aggregation defense mechanisms mitigate the threat of poisoning attacks by introducing similarity or gradient contribution evaluation to filter out malicious gradients. However, the emergence of adaptive poisoning attacks in recent years has rendered such defenses less effective. To better counter the model poisoning attacks, a hierarchical clustering-based aggregation algorithm was proposed. By processing gradients in a bottom-up manner, the algorithm enhanced the UAV swarm’s robustness against various types of poisoning attacks. Experimental evaluations on three commonly used datasets demonstrated the effectiveness of the proposed method across different attack scenarios. Compared with existing approaches, the proposed method improved the average defense success rate by 11.25% and increased the model accuracy by an average of 2.8%.

Unmanned aerial vehicle (UAV) faces the risk of data leakage and malicious attacks when collecting and transmitting data. Differential privacy technology can provide privacy protection in the communication process. However, the traditional differential privacy mechanism has the problem of poor privacy protection and data utility loss when dealing with the high dimensional or complex data in UAV communications. To solve these problems, a novel matrix differential privacy protection method named matrix Gaussian mechanism (MGM) was proposed. MGM provided privacy protection in matrix data by introducing structured noise, while using matrix covariance structure to control the direction of noise addition to minimize the data utility loss. Compared with the traditional methods, MGM could adjust the noise distribution more flexibly, improve the efficiency of privacy protection, and maintain the structural characteristics of data in the multidimensional data space. Experimental results showed that the proposed method could effectively protect data privacy with improving the efficiency of UAV communications as well as the adaptability and scalability of model training.

The rapid development of mobile network has created new development opportunities for many fields. Mobile edge computing (MEC) has attracted much attention on its widespread applications in the smart city vehicular networks due to its unique advantages. However, the increasing number of vehicular users has led to several challenges, such as network channel resource shortage, high latency, and excessive energy consumption, significantly reducing the data transmission efficiency. Therefore, an unmanned aerial vehicle (UAV)-empowered task offloading scheme was proposed for the smart city vehicular networks. By integrating the semantic communication with multiple access, multiple vehicular users could simultaneously access the network and extract semantic information for uploading to UAV, thereby enhancing the data transmission efficiency. At the same time, the game incentive mechanism was introduced to enhance the effectiveness of each participant. The simulation results showed that compared with the traditional baseline scheme, the proposed scheme could significantly improve the overall utility of the system.

The development of unmanned aerial vehicle (UAV) enabled communication faces challenges of both security and efficiency. A futuristic yet important problem faced by the current UAV-enabled communication was explored how could we make UAVs as energy efficient as possible while maintaining the communication secrecy? To solve this problem, a secrecy energy efficiency maximization (SEEM) solution for advanced dual-UAV-enabled secure communication systems was proposed. In such systems, a source UAV ensured secured information transmission with the assistance of a jammer UAV. In order to establish the optimal trade-off between security and efficiency, maximal secrecy energy efficiency (SEE) for dual-UAV enabled secure communications was seeked. Solving the issue optimally was challenging because of its non-convexity and imprecise locations of eavesdroppers. Based on the successive convex approximation (SCA) and S-procedure techniques, the SEEM solution was designed for optimizing the trajectory of source and jammer UAVs in this situation. The simulation results demonstrated that the proposed solution could establish a substantially greater SEE than other benchmark solutions for dual-UAV-enabled secure communication systems. Consequently, the proposed solution could focus on both transmission covertness and energy efficiency for dual-UAV-enabled secure communications.

With the rapid development of the low-altitude economy, the construction of communication, navigation, and surveillance infrastructure is accelerating, making the assurance of safe and orderly low-altitude operations a critical task. As the core technical support of the low-altitude supervision and management, the security of low-altitude surveillance networks directly affects the operational efficiency and stability of low-altitude airspace. It is essential to conduct an in-depth analysis of the security of the low-altitude surveillance networks, identifying potential risks and providing guidance for future system design and construction. Firstly, the current development situation of low-altitude surveillance networks was introduced, revealing the challenges in the process of low-altitude management and technology implementation. Secondly, the equipment and data assets for cooperative target surveillance and 5G-A (5G-Advanced)-based non-cooperative target perception architectures were identified, and the current research progress on cybersecurity attacks targeting these assets both domestically and internationally was discussed. Subsequently, a detailed analysis of the potential security threats posed by data security issues to low-altitude surveillance capabilities was presented, followed by an exploration of the future research directions in low-altitude surveillance security.

When automatically detecting defects in insulator images by unmanned aerial vehicle (UAV) , the frequent use of the attention module by the detection algorithm leads to large model parameters and poor real-time performance. In order to realize lightweight and high-accuracy UAV intelligent inspection, a vision transformer (ViT) based UAV insulator defect detection model based on DEtection TRansformer (UID-DETR) was proposed. Firstly, the proposed fast re-parameterization module (FREP) utilized the partial convolution (PConv) and re-parameterization convolution (RepConv) to reduce redundant computations and extract spatial features efficiently. Secondly, the efficient intra-scale interaction module (EISI) was designed for enhancing the interaction of high-level features. Thirdly, the complementary integration of high-level and low-level semantic information was realized by the feature fusion strategy of static fusion (STF) and dynamic fusion (DYF). Extensive experimental results verified the effectiveness of the proposed method on open-source synthetic foggy insulator dataset (SFID) and self-made insulator dataset (SID).