Danyang Qin

In urban cellular networks, unmanned aerial vehicles (UAVs) serve as critical aerial nodes and are extensively deployed. However, during mission execution, they are susceptible to co-channel interference (CCI) with terrestrial networks, which can adversely impact quality of service (QoS). This paper develops an optimization model for uplink communication links in spectrum-sharing cellular networks involving UAVs and proposes a joint optimization framework for UAV trajectory planning and transmission power control. Specifically, we introduce a Double and Dueling Deep Q-Network via Inverse Reinforcement Learning (D3QN-IRL) algorithm featuring a prioritized experience replay mechanism with replay time, designed to investigate the impact of varying ground user density on optimization problems. The proposed algorithm integrates expert demonstrations with Deep Reinforcement Learning (DRL), guiding UAVs to avoid interference hotspots within dynamic jamming environments and autonomously reduce transmission power when approaching densely populated user areas. Through this process, the UAV is capable of progressively learning the optimal policy to achieve joint optimization of trajectory and power allocation. Simulation results demonstrate that the proposed approach significant outperforms existing methods in terms of throughput, signal-to-interference-plus-noise ratio (SINR), and interference management, validating its effectiveness and potential for deployment in complex wireless communication environments.

For future 6G networks, wireless localization, as a radio-measurement-based method, has become a key-enabler technology for intelligent location-based services. While classical deep neural networks (DNNs) based approaches effectively improve accuracy in complex environments, their practical deployment still faces two major challenges: the labor-intensive calibration cost and the limited robustness with environmental dynamics. To address these issues, we propose SSRLoc, a self-supervised representation learning framework to achieve cross-scenario localization with minimal deployment efforts. Specifically, SSRLoc characterizes the clustering structure of unlabeled radio measurements and exploits local proximity to learn latent representations, so that the similar measurements are mapped close to each other while the dissimilar ones are separated in the representation space. In the online localization stage, the learned signal representation model is used to extract informative features by the limited radio fingerprints and the localization estimator is trained in a new scenario. Finally, the user location is inferred through a probabilistic estimation. Extensive experiments across diverse indoor scenarios show that SSRLoc achieves robust cross-scenario location estimation, providing a practical measurement methodology for low-effort and scalable wireless indoor localization.

Multi-agent collaborative perception has shown significant potential in improving environmental awareness, especially in complex and dynamic environments where it enables a more precise and comprehensive understanding. However, existing collaborative perception systems still face several challenges, particularly during feature fusion, where issues such as information redundancy and noise interference arise. To address these challenges, this paper introduces the Dual-Attention Weighting Collaborative Perception (DAW-CoPe) framework. The framework integrates dilated convolutional techniques within the Multi-Scale Spatial Attention Module (MSAM), allowing for the capture of multi-dimensional object features across different scales and the effective integration of richer contextual information, thereby enhancing the model’s perceptual capabilities. Furthermore, the incorporation of local convolutions strengthens the model’s ability to extract fine-grained features, improving its capacity to capture local details during feature fusion and reducing the risk of losing important information. To further enhance perception accuracy, we introduce a dual-attention weighting (DAW) mechanism: a same-level, learnable gating module that jointly reweights channel semantics and multi-scale spatial cues, thereby emphasizing informative channels and critical spatial regions. By suppressing redundant and noisy signals, this mechanism delivers substantial gains in performance and robustness across perception tasks. Experimental results demonstrate that DAW-CoPe outperforms existing state-of-the-art methods on three public datasets, significantly boosting the accuracy, reliability, and adaptability of collaborative perception systems. The code associated with this research will be made publicly available at https://github.com/HIT-K/DAW-CoPe

The Journal retracts the article “Indoor floor localization based on multi-intelligent sensors”

Reliable image transmission is critical for applications such as autonomous driving, collision avoidance, and real time traffic monitoring in the field of vehicles. However, extreme and dynamic vehicular environments impose significant challenges on reliable image transmission. Traditional communication methods that separately design source and channel coding fail to fully exploit the signal-to-noise ratio (SNR) information in vehicular channels and lead to suboptimal performance. Although recent deep learning (DL) based approaches have shown promising results in enhancing image transmission, they are often constrained by high storage requirements and limited adaptability to the highly variable channel conditions in vehicular environments. In this paper, a novel DL-based SNR adaptive attention joint source channel coding (SAA-JSCC) framework tailored for vehicular wireless communication systems is proposed. The proposed SAA-JSCC dynamically adjusts encoding and decoding strategies in real time by leveraging instantaneous SNR feedback, thereby optimizing transmission efficiency and reliability. A deep neural network is employed to directly map input RGB images to complex-valued baseband samples, while the receiver integrates the decoding and decision-making processes within its neural architecture. Comprehensive simulations under diverse vehicular channel conditions demonstrate that the SAA-JSCC approach significantly outperforms existing JSCC methods in terms of Peak Signal-to-Noise Ratio (PSNR) for image transmission, particularly under low SNR scenarios prevalent in vehicular environments. Furthermore, the proposed method delivers significant PSNR gain with a modest increase in FLOPs and negligible parameter growth, while remaining far smaller model size than prior deep baselines, thereby facilitating more efficient and scalable real time communication in next-generation vehicular networks.

With the development of the Internet of Things, cross-view geolocation of satellite images and uncrewed aerial vehicles (UAVs) aerial images is gaining attention due to its significant potential in denied-navigation environments under global navigation satellite system (GNSS). However, inadequate feature representation between views, as well as uncertainties in positional offsets and distance scales, constitute key challenges in this field. Existing research primarily focuses on extracting comprehensive and fine-grained feature information. However, effective feature representation and alignment are equally critical. Therefore, a novel visual localization algorithm based on multiscale feature fusion based on improved transformer architecture (MFFT) is proposed. The algorithm combines Adaptive semantic Feature extraction based on Improved ASPP (AFIA) model and spatial pyramid based on multiscale subsampling (SPMS) model, and has strong robustness. MFFT cooperatively uses AFIA and SPMS to successfully achieve an effective balance between feature representation and alignment. Experimental results show that, on the common benchmark data set DenseUAV, compared with the most advanced algorithms in the same field, the R@1 of MFFT algorithm is improved by at least 2.23% and the SDM@1 is improved by at least 1.64%, thus achieving the most advanced cross-view matching performance, verifying that the proposed algorithm has strong competitiveness in UAV visual self-localization tasks, and opening up a new technical path for future UAV autonomous navigation and diversified mission execution in complex environments.