With the development of the sixth-generation (6G) communication networks, the scale of internet of things (IoT) devices is rapidly expanding. However, the limited computational and energy resources have become the major bottlenecks constraining IoT devices in processing complex tasks and providing high-quality services. In order to solve this problem, a novel unmanned aerial vehicle (UAV)-mounted reconfigurable intelligent surfaces (RIS) assisted mobile edge computing (MEC) with backscatter system is proposed. In this paper, a system energy efficient (EE) maximization problem is formulated by jointly optimizing the reflection coefficients, computational resources, time allocation, RIS phase shifts, and UAV trajectories while satisfying the task constraints, energy causality constraints, and trajectory constraints. To solve the established non-convex optimization problem, a three-stage alternating optimization algorithm is developed based on the Dinkelbach algorithm. The efficient solution of each subproblem is realized by leveraging the Lagrangian dual method, combined with the sub-gradient descent and successive convex approximation techniques. Furthermore, the closed-form expressions for the CPU computation frequency, backscatter reflection coefficient, and RIS phase-shift coefficients are derived. Simulation results show that the proposed scheme achieves superior system EE compared with the benchmark and simplified schemes.

Precise positioning and identification of unauthorized unmanned aerial vehicles (UAVs) are of crucial importance for spectrum security and privacy protection in future intelligent networks. Although various single-modality approaches have been investigated, their performance degrades under the sensor-specific noise, resulting in suboptimal performance and robustness. To address these security challenges, we propose a multi-layer radio frequency (RF)-vision fusion framework that synergistically exploits temporal-spectral features of UAV RF signals and spatial-visual information to achieve precise and robust UAV positioning and identification. Moreover, a corresponding unified RF-Vision fusion Network (RFViNet) is designed to exploit the RF-vision cross-modal complementary and semantic synergy. Specifically, by leveraging the novel RFinformed proposal generation, RF-enhanced feature modulation, and RF-guided semantic query modules, the RFViNet effectively exploits the complementary strengths of RF and visual modalities. Furthermore, a practical RF–vision platform is developed to evaluate the performance of our method under various challenging conditions. Experimental results on the real-world dataset demonstrate that the proposed method achieves a competitive 85.8% average precision AP50, highlighting its potential for enhancing the spectrum security in future intelligent wireless networks.

The rapid proliferation of UAV technology in civilian and military sectors has brought significant benefits but also raised concerns about unauthorized unmanned aerial vehicle (UAV), which pose potential risks to public safety. Current anti-UAV systems, which rely on radar, acoustic, antenna or visual sensors, encounter specific challenges, such as limited positioning ranges, susceptibility to noise, background interference and adverse environmental conditions. To address these limitations, we propose an RF-visual fusion-based memory augmentation network (RVUAV-Net) that integrates RF and visual images, enhancing the UAV positioning accuracy and robustness by utilizing the motion patterns derived from the historical UAV trajectory. Our proposed method capitalizes on the spatial-temporal characteristics of the historical data, enabling precise UAV positioning during partial occlusion and effective responses to high-speed movements. Moreover, our proposed RVUAV-Net minimizes false alarms in the complex environments by distinguishing UAVs from similar flying objects. Experimental results demonstrate superior positioning performance of our proposed method, highlighting its potential for anti-UAV in real-world scenarios where accuracy and continuity are critical.

Precise unmanned aerial vehicle (UAV) detection over long distances is of crucial importance for guaranteeing the airspace security. Although deep learning-based vision detectors have been developed, they still rely on a large amount of hand-crafted fixed feature priors. The existing static dense-based detectors suffer from the severe mismatch and imbalance between the small size and the high mobility of UAVs. To solve the problem, a novel multimodal fusion-based dynamic sparse UAV detection framework is proposed. The framework reformulates the feature priors in a completely dynamic sparse paradigm by using the radar data. Based on the framework, a vision-radar fusion-based dynamic sparse network (Vira-DSNet) is proposed for more balanced and robust UAV detection. The Vira-DSNet exploits our designed dynamic sparse candidate generator and radar-guided semantic feature transform to generate a small set of customized high-quality object candidates and semantic features based on the radar data. Moreover, based on Hungarian bisection matching, our Vira-DSNet eliminates the post-processing and is completely end-to-end differentiable. Furthermore, the Vira-DSNet is deployed in our developed actual vision-radar fusionbased UAV detection system to evaluate the performance in the practical applications. Experimental results demonstrate that our Vira-DSNet achieves an average precision AP50of 88.2%. It is also shown that the average recall AR1of Vira-DSNet is higher than the state-of-the-art scheme by 10.1%, while maintaining the real-time performance.

Uncrewed aerial vehicles (UAVs) positioning is of crucial importance in diverse applications. However, it is extremely challenging to realize the precise UAVs positioning over long distances due to the small size and dramatic scale variations associated with the high mobility in the wide area. To tackle this issue, a multimodal scale normalization framework is proposed for the scale-robust precise pixel-level UAV positioning. The framework exploits our proposed distance-aware image slicing and distance-aware scale normalization module. Moreover, a modal fusion-based scale normalization network is proposed that can accept arbitrary low-resolution UAV patches and produce the consistent high-resolution images at a uniform UAV instance scale with a single learnable model. The proposed framework is generic and can be directly used in the existing pixel-level positioning pipelines to improve the positioning performance and scale robustness. To verify the proposed framework in the real application, a practical vision-radar UAV positioning system is developed. Experimental results on the real-world dataset demonstrate the generality and effectiveness of our framework. Moreover, the ablation experiments also confirm the contribution of each module in the framework.

Anti-unmanned aerial vehicle (UAV) systems are crucial for preventing unauthorized individuals from exploiting UAVs for illegal activities, including surveillance and attacks. Precise real-time positioning of small UAVs is the premise of the effective operation of anti-UAV systems. However, its performance is confined due to the small size of the target and its high susceptibility to disturbance caused by birds or other aircraft. To tackle this problem, a radio-frequency (RF)-visual directional fusion framework is proposed for precise UAV positioning. In the framework, radio signals are aligned with images by jointly calibrating the array antenna and camera. The spatial spectrum is extracted by an array antenna to concentrate on target areas within the image modal. Moreover, in order to improve the precision of joint calibration, a segmentation-based denoising method is proposed to remove the spectrum noise. Furthermore, a practical anti-UAV positioning platform is established, and two synchronized data sets, which include visual images and UAV RF signals, are collected on the platform. Experimental results demonstrate that our proposed framework improves positioning accuracy and robustness compared to the benchmark methods.