Self-supervised monocular depth estimation leverages photometric consistency across views as supervisory signals, yet inherently suffers from compromised geometric fidelity at object boundaries, surface smoothness, and cross-view consistency due to 3D structural information loss during 2D projection. This paper introduces Diff3D-Net, a novel framework integrating multi-level 3D geometric constraints to overcome these limitations. Key innovations include: a fixed-parameter RPM-Net point cloud registration module generating geometrically grounded supervision; a pose consistency constraint that directly optimizes the pose network using RPM-Net-refined transformations to suppress error accumulation; a dual-path photometric constraint adaptively fusing reconstructions from original and optimized poses to enhance low-texture robustness; and a multi-view 3D structural consistency constraint enforcing global point alignment via minimum residual selection. Evaluations on KITTI and DDAD show Diff3D-Net's state-of-the-art performance, achieving 0.094/4.220 Abs. Rel/RMSE on KITTI. The model also exhibits strong cross-domain generalization in zero-shot settings, outperforming competitors on Cityscapes, NYU Depth V2 and DrivingStereo datasets without fine-tuning, highlighting its IoT application potential. Despite higher training costs, its inference achieves an optimal speed-accuracy trade-off, with a lightweight variant reaching 51.73 FPS -sufficient to enable real-time applications such as UAV navigation on resource-limited IoT edge devices. Ablations confirm each constraint's contribution, and qualitative results demonstrate robustness in dynamic scenes and complex geometries.

Visible Light Positioning (VLP) has emerged as a promising indoor localization technology owing to its high accuracy, low power consumption, and lighting compatibility. The Received Signal Strength (RSS)-based multi-anchor VLP pre-serves these advantages while having drawn considerable research attention due to its simple implementation and high reliability. However, its large-scale deployment encounters challenges at every stage: inefficient anchor calibration, limited model-based ranging performance, and robustness reduction from undetected gross errors. To address these issues, we propose a VLP and inertial navigation system fusion framework comprising an anchor position estimation module, a distance estimation module, and a fusion positioning module. For anchor calibration, a LiDAR-inertial odometry-based calibration scheme enhanced by a twolayer optimization strategy is introduced, which provides prior knowledge of anchor positions for the whole system. To improve the performance of the RSS-based ranging method, a hybrid Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN)-Bidirectional Long Short-Term Memory (Bi-LSTM) network with an embedded distance quality assessor is developed, achieving over 50% higher accuracy than model-based baselines. It delivers precise distance estimates and their validity labels for measurement updates in subsequent fusion positioning. Additionally, a two-stage error detection mechanism filters low quality observations by combining network-generated usability labels with prior-state estimates. The system consistently attains decimeter-level positioning across various trajectories, meeting the needs of diverse Internet of Things applications.

Monocular visual-inertial SLAM (VINS) confronts formidable challenges in subterranean environments such as mines, where inadequate illumination and feature-deficient textures substantially degrade system performance. Conventional VINS frameworks and contemporary line-enhanced SLAM methodologies exhibit significant performance deterioration under such conditions. This paper presents RPLF-VINS, a robust monocular VINS framework integrating hybrid point-line flow features, specifically engineered for operation in photometrically challenging underground environments. Our system introduces two principal innovations: a SSR-CLAHE image enhancement pipeline synergizing Single-Scale Retinex (SSR) with Contrast-Limited Adaptive Histogram Equalization (CLAHE) to address non-uniform illumination while preserving critical texture details through global luminance correction and localized contrast optimization; and a novel line-feature residual estimation strategy that combines Euclidean distance with plane-normal-angle metrics to more accurately model 3D reprojection errors of line observations, balancing geometric and angular error contributions. Extensive experimental validation across public benchmarks and custom underground datasets demonstrates our system’s superior performance, with quantitative evaluations revealing 40.0%, 50.0%, and 25.0% RMSE reductions compared to state-of-the-art VINS-Mono, PL-VINS, and EPLF-VINS implementations, respectively. RPLF-VINS therefore offers practical value for underground IoT applications by enabling more reliable navigation and mapping in low-light, low-texture settings.

Visible Light Positioning (VLP) has emerged as a promising indoor localization technology due to its high accuracy, low cost. However, it still faces challenges such as environmental interference, signal noise, and occlusion. To address the above issues, a Bidirectional Encoder Representation from Transformer (BERT)-enhanced VLP and inertial navigation fusion positioning system is developed. Firstly, to tackle the problem of inaccurate ranging caused by signal noise, we propose a Transformer-based network, VLP-BERT, which leverages long-sequence masking to enhance the network’s feature extraction capabilities from visible light signals. Moreover, the VLP-BERT is integrated into an autoencoder-decoder architecture for signal denoising. Secondly, to overcome the limitations of traditional ranging models in complex environments, a deep learning-based centralized VLP ranging model is proposed. Finally, to enhance the system’s reliability under varying conditions, a tightly coupled fusion method integrating VLP with Pedestrian Dead Reckoning (PDR) is proposed, incorporating error detection and state-constrained strategies. Extensive experimental evaluations demonstrate the effectiveness of VLP-BERT in both denoising and accurate ranging. The system was compared with nine different methods, the results show that the proposed tightly coupled approach not only achieves sub-meter-level accuracy but also significantly enhances the system’s robustness, even in challenging scenarios such as signal blockage and poor signal quality.

Specific emitter identification (SEI) is an evolving methodology aimed at discerning individual sources by extracting the radio frequency fingerprint (RFF) inherent within signals. This study confronts the practical challenges associated with digital predistortion (DPD) techniques and insufficient data impacting SEI performance. A novel model is proposed for few-shot SEI (FS-SEI), integrating convolutional neural network (CNN) embedding and metric learning, meticulously tailored to scenarios constrained by limited sample availability. The investigation rigorously examines the ramifications of DPD techniques on SEI, accentuating their potential to erode the salient characteristics defining emitter identities. In the context of SEI with few samples affected by DPD, our methodology begins by representing signals as time-frequency images, followed by the extraction of deep features via a CNN architecture. These extracted features are subsequently embedded within a relational network to assess interfeature relationships. Experimental results substantiate the deleterious impact of DPD techniques on SEI, highlighting the superiority of our proposed approach over traditional CNNs, exhibiting significantly enhanced accuracy in limited sample scenarios.

As a critical topic of Internet of Things applications, smartphone-based indoor navigation has a rapidly growing need in various applications. However, indoor navigation technology is unreliable when facing a challenge in complex indoor environments. This article presents a tightly coupled (TC) integration of pedestrian dead reckoning (PDR) and Bluetooth for indoor pedestrian navigation and enhances it from three approaches. We first establish a Gaussian-based distance model (GDM) that improves the signal path-loss model to incorporate the prior information on the variation of signal volatility with distance. Then, the use of map information and a back-off strategy to optimize the particle transfer strategy further improves the positioning accuracy and rationality of the system. Moreover, we leverage behavioral landmarks, signal landmarks, and distance information to build a graph optimization model to optimize the proposed navigator. We have extensively verified the proposed navigator and compared it with the existing solutions and systems. Experimental results demonstrated that the average errors of the proposed solutions in three scenes were 34.71% of Bluetooth, 14.04% of PDR, 45.13% of the extended Kalman filter, 57.83% of the unscented Kalman filter, and 56.10% of PF, respectively. The results showed that our proposed solution has apparent advantages, especially when addressing the issues of incorrect trajectory updating and divergence of the system in a complex environment.

Indoor localization using embedded multi-sensors has attracted considerable attention over the past decades. However, traditional localization solutions usually either depend heavily on elaborate scenes with infrastructure layout or provide inertial-based results with severely degraded accuracy in a long-term positioning environment. Multi-story complex buildings require accurate indoor positioning results, including height information, while previous studies rarely involved indoor 3D location research. This paper proposes an accurate indoor 3D location algorithm based on MEMS/vision by using a smartphone. We make full use of both visual and inertial modal information. Visual positioning is used to achieve global positioning, while relative positioning is based on traditional PDR. First, the MEMS-based relative positioning solution is proposed to determine 3D position by using complementary filter. Secondly, efficient image-based localization (IBL) algorithm is employed, and we use the prior attitude information to obtain the geometric constraint for the global map and improve efficiency of 2D-3D correspondence. Moreover, we apply a dynamic fusion strategy to integrate the MEMS-based results and visual-based locations, providing the accurate 3D position continuously by using relative and global solutions. Experiments conducted in typical indoor scenes showed that the proposed localization system can obtain decimeter 3D position most time (over 85% 3D location errors are within 0.45 m), which effectively mitigates the cumulative errors of low-cost IMU.

Abstract Cyclostationary feature detection is one of the widely used spectrum sensing techniques. Its greatest advantage is that it can effectively separate the signal from the noise under the condition of a low signal‐noise ratio (SNR). Although cyclostationary feature detection is not affected by noise uncertainty like energy detection (ED), conventional cyclic stationary feature detection needs to know the prior conditions of signal cycle frequency (CF), cycle period, and so on. Moreover, the communication signals of authorized users are generally weak, which greatly affects the detection efficiency of cyclic stationary features. Therefore, this paper proposes a blind detection technique of cyclostationary characteristics based on conventional beam synthesis multi‐antenna. The principle is to enhance the signal receiving intensity of the antenna by using beamforming technology, and realize the blind detection of cyclostationary signals by forming the direction of arrival angle. Not only that, the authors also make full use of the Wilk's approximation theorem and the generalized likelihood ratio test (GLRT) to derive the detection probability and false alarm probability. The simulation results show that the proposed cyclic stationary signal detector based on beam synthesis multi‐antenna can realize blind detection of unknown cycle signals in a low SNR environment.

In the adaptive traffic signal control (ATSC), reinforcement learning (RL) is a frontier research hotspot, combined with deep neural networks to further enhance its learning ability. The distributed multiagent RL (MARL) can avoid this kind of problem by observing some areas of each local RL in the complex plane traffic area. However, due to the limited communication capabilities between each agent, the environment becomes partially visible. This paper proposes multiagent reinforcement learning based on cooperative game (CG‐MARL) to design the intersection as an agent structure. The method considers not only the communication and coordination between agents but also the game between agents. Each agent observes its own area to learn the RL strategy and value function, then concentrates the Q function from different agents through a hybrid network, and finally forms its own final Q function in the entire large‐scale transportation network. The results show that the proposed method is superior to the traditional control method.

Flow Mapping, also known as spatial interaction data visualization, has become widely used for exploratory spatiotemporal data analysis to understand complex spatial phenomena such as human migration, commercial trading, and social networks. The unitary flow mapping architecture, in which data storing, computing and representation are deployed in a single computer, is facing the challenges being brought from increasing data scale, higher timing demand, computing complexity of visual clutter detecting and integrating with other GIS and spatio-temporal analysis tools. In this paper, a novel 3-tiers flow mapping service architecture is proposed. In this architecture, flow data integration tier provide a unified data access interface for variant data sources; flow mapping models tier provide a computing resource pool to support different flow mapping algorithms and scalable computing capability; and result visualization tier to view map interactively. In this paper, we expand the OGC Web Map Services (WMS) standard protocol to support spatio-temporal interaction data visualization and analytics, and integrate WMS-based flow mapping service with other map resources by JavaScript toolkits in browsers. This architecture is validate to be improved in performance and scalability by three typical application cases.