Unmanned aerial vehicles (UAVs), relying on wireless communication, are inevitably influenced by the complex electromagnetic environment, attributed to the development of wireless communication technology. The modulation information of signals can assist in identifying device information and interference in the environment, which is significant for UAV communication environment monitoring. Therefore, in scenarios involving the communication of UAVs, it is necessary to find out how to perform the spectrum monitoring method to obtain the modulation information. Most existing methods are unsuitable for scenarios where multiple signals appear in the same spectrum sequence or do not use an end-to-end structure. Firstly, we established a spectrum dataset to simulate the UAV communication environment and developed a label method. Then, detection networks were employed to extract the presence and location information of signals in the spectrum. Finally, decision-level fusion was used to combine the output results of multiple nodes. Five modulation types, including ASK, FSK, 16QAM, DSB-SC, and SSB, were used to simulate different signal sources in the communication environment. Accuracy, recall, and F1 score were used as evaluation metrics. The networks were tested at different signal-to-noise ratios (SNRs). Among the different modulation types, FSK exhibits the most stable recognition performance across different models. The proposed method is of great significance for wireless radio spectrum monitoring in complex electromagnetic environments and is adaptable to scenarios where multiple receivers are used in vast terrains, providing a deep learning-based approach to radio monitoring solutions for UAV communication.

Multi-unmanned aerial vehicle (UAV) systems are crucial for establishing resilient communication networks in disaster-stricken areas, but their limited energy and dynamic characteristics pose significant challenges for sustained and reliable service provision. Optimizing resource allocation in this situation is a complex sequential decision-making problem, which is naturally suitable for multi-agent reinforcement learning (MARL). However, the most advanced MARL methods (e.g., multi-agent proximal policy optimization (MAPPO)) often encounter difficulties in the “loosely coupled” multi-UAV environment due to their overly centralized evaluation mechanism, resulting in unclear credit assignment and inhibiting personalized optimization. To overcome this, we propose a novel hierarchical framework supported by MAPPO with decoupled critics (MAPPO-DC). Our framework employs an efficient clustering algorithm for user association in the upper layer, while MAPPO-DC is used in the lower layer to enable each UAV to learn customized trajectories and power control strategies. MAPPO-DC achieves a complex balance between global coordination and personalized exploration by redesigning the update rules of the critic network, allowing for precise and personalized credit assignment in a loosely coupled environment. In addition, we designed a composite reward function to guide the learning process towards the goal of proportional fairness. The simulation results show that our proposed MAPPO-DC outperforms existing baselines, including independent proximal policy optimization (IPPO) and standard MAPPO, in terms of communication performance and sample efficiency, validating the effectiveness of our tailored MARL architecture for the task. Through model robustness experiments, we have verified that our proposed MAPPO-DC still has certain advantages in strongly coupled environments.

With the rapid development of in-vehicle electronic technology and artificial intelligence, Internet of Vehicles (IoV) technology, as an effective integration of the two, greatly reduces the probability of road traffic accidents. However, the current IoV system is not perfect for the control process of abnormal vehicles. Therefore, in order to strengthen the management and control of abnormal vehicles in the IoV, it is extremely necessary to propose a method for interfering with IoV signals. Among the current popular intelligent interference methods, most of them rely on the prior knowledge of the signal to deduce the best interference waveform. However, These methods rely too much on prior knowledge to be practically applicable. Therefore, in view of the shortcomings of the current communication interference technology, this study proposes an interference waveform generation technology based on convolutional autoencoders. The convolutional autoencoder was used to change the features on the fully connected layer to generate an interference waveform that is very similar to the received signal waveform, and the interference waveform is sent to the receiver to realize the control of the IoV signal. The simulation results show that the interference waveform generation technology proposed in this study can make the bit error rate (BER) reach 38.4\% within the signal-to-interference ratio (SIR) from -10dB to -15dB.

Inferring the topology of Wireless Communication Networks (WCNs) from radiated electromagnetic signals is crucial for network layer radio management. This study analyzes WCN Topology Inference (TPI) mechanism and derives that the topology information between two nodes is distributed across their own signals and those of neighboring nodes. Existing WCN-TPI methods typically utilize a fraction of the topological information from two-node signals, limiting their performance. To address this challenge, this study proposes a multi-node TPI method, highlighting its advantages in terms of interference signal, sequential feature, and information quantity. Additionally, the lower bound of the exact probability for TPI is determined using Acknowledge feature, whereas the impact of factors such as node degree, Signal-to-Noise Ratio (SNR), sampling rate, and time slot length on TPI is explored. Finally, the proposed multi-node TPI method implemented using Graph Convolutional Network (GCN). Experimental results show that the GCN-based approach outperforms existing methods in both accuracy and efficiency. It achieves 0.978 accuracy at 0 dB and maintains the highest accuracy and precision under 40% data loss over a 5.7 s signal, demonstrating strong robustness. The accuracy degrades by less than 0.02 under multipath conditions. Moreover, on a 30-node network, it runs 0.04 s faster than the Transformer method with comparable accuracy.