This paper presents a novel semantic communication system for efficient and resilient control signal transmission in unmanned aerial vehicles (UAVs). Traditional bit-level transmission methods face challenges under poor channel conditions and dynamic environments, where packet loss and the high dimensionality of control signals impact operational stability. To address these challenges, we propose a long short-term memory (LSTM) based semantic encoding and decoding framework that compresses essential control information. Additionally, a trajectory prediction model is integrated into the semantic framework to refine control signals, thereby enhancing accuracy. Our design includes an encoder at the ground control station (GCS) and a lightweight decoder onboard the UAV, making it suitable for resource-constrained UAVs without onboard control signal processing. Performance evaluations across varying channel conditions demonstrate that the proposed semantic communication method reduces trajectory mean squared error by over 60% and saves communication costs by at least 37% compared to traditional bit-level communication approaches.

Achieving ultra-reliable and low-latency communication (URLLC) in integrated terrestrial networks (TNs) and non-terrestrial networks (NTNs) is essential for the advancement of 6G networks. However, there is a significant lack of theoretical tools for analyzing probabilistic end-to-end (E2E) delay in such integrated networks. This analysis is crucial for estimating URLLC delay threshold violations and ensuring compliance with URLLC delay requirements. To address this gap, this paper introduces a systematic approach based on stochastic network calculus (SNC) utilizing moment-generating functions (MGFs) to compute the E2E delay violation probability for target traffic in unmanned aerial vehicles (UAVs) assisted integrated TNs and NTNs. The unique aspects of this scenario include two components: (1) TNs using conventional ground base stations, and (2) NTNs employing UAVs to relay data to ground base stations or high-altitude platform stations (HAPS). Our approach enhances the calculation of MGFs for service processes by considering transmission failures and blockages over sub-6 GHz and mmWave fading channels. This improvement offers a more precise performance characterization and facilitates the derivation of a closed-form expression for the upper-bounded statistical delay violation probability. Numerical results confirm that UAVs are advantageous in ensuring E2E delay guarantees by facilitating line-of-sight (LOS) transmission. Additionally, we provide insightful observations for resource allocation strategies in UAV-assisted integrated TNs and NTNs.

This demo presents a novel implementation of physical layer key generation (PLKG) in UAV communication systems by integrating deep learning for key reconstruction. Secure communication is a significant challenge in UAV networks due to their high mobility, frequent topology changes, and vulnerability to eavesdropping. Traditional cryptographic methods are inefficient in such dynamic environments because of high computational costs and latency. Our approach leverages multimodal learning to enhance resilience against Doppler effects and dynamic channel variations. The demonstration showcases a working prototype that extracts channel state information (CSI), predicts UAV trajectory, and reconstructs cryptographic keys with improved consistency and lower mismatch rates compared to conventional methods. Our system utilizes ESP32 microcontrollers for real-time CSI acquisition and Raspberry Pi 4 for deep learning-based processing. We provide a graphical user interface (GUI) that visualizes real-time CSI fluctuations and reconstructed key bits, demonstrating the framework’s resilience to dynamic channel variations.

This paper introduces the first implementation of physical layer key generation (PLKG) on a real-world unmanned aerial vehicle (UAV) platform. To tackle the unique challenges of high mobility and dynamic communication environments in Internet of Drones (IoD) networks, we propose a novel multimodal learning framework for enhancing PLKG in UAV-to-ground communications. Static channel state information (CSI) features and dynamic UAV trajectory data are extracted using convolutional neural networks (CNN) and long short-term memory (LSTM) networks, respectively. Leveraging conditional embedding techniques, the predicted UAV position and velocity are integrated into the input space of the key reconstruction network as conditional features. The trained network serves as a Doppler-resilient feature extraction mapping function, thus achieving robust and consistent key generation under varying mobility conditions. Experimental results demonstrate lower key mismatch rates and higher reliability compared to existing CSI-based key generation methods.

Accurate and reliable localization is a prerequisite for unmanned aerial vehicle (UAV) swarm applications. However, conventional GPS or RF-based localization systems often do not function effectively in highly dynamic and unstable mobile ad-hoc environments. This paper proposes a new approach to localize UAVs accurately in unknown communication environments with anomalous GPS reception, based on the received signal strength (RSS) between UAVs. The proposed approach is non-trivial, given the combinational nature of the considered problem and the requirement of high localization accuracy in the UAV application scenario. The key idea of the proposed approach is to solve the position mapping problem by refining a convex relaxation formulation that considers whether the target to be localized is inside or outside the convex hull formed by the anchors. In addition, a variational graph autoencoder is utilized to learn the latent representations for the undirected graph formed from the estimated position, which is then used to calculate the anomaly score. The optimal anchor node selection is obtained by solving a fractional knapsack problem that takes into account the anomaly score of different anchor combinations. Simulation results demonstrate that the proposed approach achieves higher detection and localization accuracy and is more robust to RSS measurement errors compared to the baseline schemes.