This study presents a modular smart lighting system that employs an open-loop, daylight-based control strategy supported by artificial intelligence (AI). Unlike conventional closed-loop systems that rely on multiple indoor sensors and complex wiring, the proposed system adopts a distributed architecture in which a Master LED Luminaire (MLL) coordinates Slave LED Luminaires (SLLs). The MLL incorporates a Support Vector Regression (SVR)-based daylight prediction model that estimates indoor daylight availability using fixed architectural parameters such as room dimensions, window geometry, and surface reflectance together with real-time outdoor irradiance data. The system was implemented and validated in a public building under real operating conditions. Comparative evaluation with traditional on–off control and sensor-based closed-loop control indicated that, while the closed-loop approach achieved the lowest artificial lighting demand (56.4%), the proposed system required only slightly more (66.4%) yet offered clear advantages in scalability, ease of installation, and reduced infrastructure costs by eliminating the need for multiple sensors. Integration with IoT-enabled communication further allows real-time parameter updates and adaptive dimming control. The findings demonstrate that the proposed method provides a scalable and cost-effective retrofit solution, addressing the installation, maintenance, and adaptability challenges of conventional lighting control systems.

The Internet of Vehicles (IoV) marks a revolutionary leap in transportation, integrating vehicles with the Internet to enhance convenience, intelligence, and efficiency. As IoV applications continue to expand, a spectrum of challenges and untapped opportunities emerge. Among these, the security of IoV stands out as a critical issue in transportation systems, given its direct impact on road safety. To this end, this survey conducts an in-depth analysis of IoV security challenges and introduces a distinctive approach by systematically categorizing threats into inside-vehicle and outside-vehicle domains, providing a comprehensive understanding of the full spectrum of risks. Meanwhile, we examine defense mechanisms designed to counter these threats and explore proactive strategies to enhance IoV security. In addition, this study explores emerging protection techniques, such as AI-driven intrusion detection, blockchain-based trust management, and 5G-enabled secure routing, demonstrating how these technologies can be effectively integrated to safeguard IoV systems. Furthermore, this survey provides actionable recommendations and forward-looking research directions to guide the development of real-world implementations, standardized procedures, and regulatory frameworks, benefiting stakeholders, policymakers, and researchers. As IoV continues its rapid evolution, these insights offer a comprehensive roadmap to strengthen IoV security and contribute to a safer, more resilient intelligent transportation ecosystem.

Road traffic data imputation is an essential component of Intelligent Transportation Systems (ITS). However, the spatio-temporal characteristics in traffic data are complex and diverse, and existing methods are unable to comprehensively extract them. This article, through in-depth observations of traffic system dynamics, innovatively taxonomizes the complex spatio-temporal dependencies into four key dimensions: Geographical Spatial Correlations (GSC), Latent Spatial Correlations (LSC), Intra-sensor Temporal Correlations (ITC), and Cross-sensor Temporal Correlations (CTC). Motivated by this taxonomy, we aim at constructing a novel framework that holistically utilizes these spatio-temporal correlations to improve imputation accuracy. Subsequently, we propose a Multi-View Spatio-Temporal Correlation Awareness Network (MVSTA) for traffic data imputation, which incorporates two specifically designed modules: a Unified Spatial Correlation Awareness module (USCA) and a Collaborative Temporal Correlation Awareness module (CTCA). The USCA integrates GSC and LSC into a unified representation by jointly modeling physical proximity and data-driven dependencies. The CTCA collaboratively extracts ITC and CTC by capturing both local temporal patterns in individual sensors and interactive dynamics across different sensors. Extensive experiments on three real-world traffic datasets demonstrate that MVSTA significantly outperforms all baselines, validating the effectiveness of our proposed taxonomy and framework.

WiFi-based human activity recognition (HAR) holds significant promise for ubiquitous sensing in smart environments. A critical challenge is enabling systems to dynamically adapt to evolving scenarios, learn new activities without catastrophically forgetting prior knowledge, and meet edge devices’ computational constraints. Current approaches struggle to reconcile these due to high historical data storage demands and inefficient parameter utilization. We propose WECAR, an end-edge collaborative inference and training framework for WiFi-based continuous HAR. In this framework, edge devices handle model training, lightweight optimization, and updates, while end devices perform efficient inference. WECAR introduces two key innovations, i.e., dynamic continual learning with parameter efficiency and hierarchical distillation for end deployment. For the former, we propose a transformer-based architecture enhanced by task-specific dynamic model expansion and stability-aware selective retraining. For the latter, we propose a dual-phase distillation mechanism that includes multi-head self-attention relation distillation and prefix relation distillation. We implement WECAR based on heterogeneous hardware using Jetson Nano as edge devices and the ESP32 as end devices, respectively. Our experiments across three public WiFi datasets reveal that WECAR not only outperforms several state-of-the-art methods in performance and parameter efficiency, but also achieves a substantial reduction in the model’s parameter count post-optimization without sacrificing accuracy.

Radio Frequency (RF) has been widely studied, but multiple rechargeable devices may block each other, causing the mutual interference and reducing the charging utility. In recent years, pendant charging has been proposed. In pendant charging, the charger is suspended vertically downward, and the rechargeable devices are placed on the ground, thus avoiding the mutual interference. However, the existing studies only consider the pendant charging with fixed heights, where the usability is limited. In this paper, we study the height-adjustable directional pendant wireless charging system, where the directional wireless chargers are suspended vertically, and the charging ranges and received powers can be changed by adjusting the heights of chargers. We formulate the Utility Maximization Height-adjustable charging Scheduling (UMHS) problem to maximize the total utility of the wireless charging system. We first propose a height discretization method to reduce the search space significantly and obtain a discretized problem with approximation ratio to the UMHS problem. Then, we transform the discretized problem into an equivalent Multi-Dimensional Knapsack Problem (MDKP). Finally, an approximation algorithm for UMHS is proposed through the height discretization and the approximation algorithm for MDKP. The results show that our algorithm can increase up to 24.41% and 18.67% total utility compared with benchmark algorithms in extensive simulations and field experiments, respectively.

The security of underwater sensor networks (USNs) is challenging in ocean research. The primary objective of this work is to investigate the susceptibility of USNs to passive attacks, such as source location privacy (SLP). A trusted Sybil node-based source location scheme is introduced in this manuscript to fulfil the above purpose. The Sybil nodes and suitable fake source nodes are initially inserted into a network. The main intention is to generate false source data by using fake source nodes generated with evidence theory for data fusion, thereby disguising the traffic carried by the source information. After that, separate transmission windows were scheduled for the real and false packets to prevent data conflicts. The Sybil nodes are also used during the multi-path data routing process, which occurs from the source node to the destination node. In addition to diversifying the routes, it raises the bar for the adversary's ability to track the source information's route while it is being sent without conflict. The proposed TSN-SLP scheme outperformed SLPRRFPR, EECOR, and ARR by 21%, 53%, 50%, and 17.4%, respectively, in average packet data rate, safety time, transmission delay, and energy use, according to the experimental results.

Connectivity restoration is essential for ensuring continuous operation in wireless sensor networks (WSNs). However, existing works lack enough network robustness when suffering from the secondary external damages. In this article, we propose a novel connectivity restoration scheme to address this problem. This scheme comprises three connectivity mechanisms regarding relay segment selection in different regions. The first one is a data traffic decentralization mechanism, which establishes more transmission paths near the sink for reliability improvement and traffic load balancing. The second one is a segment shape selection mechanism, in which the segments with high-reliability preferably become the relay segments for greater network robustness. The third one is a traffic load transfer mechanism, in which data traffic is transferred from a high-load segment to a low-load segment for balancing energy depletion of the network. The distinctive characteristics of this work are twofold: different regions perform diverse connectivity restoration approaches according to the demand diversity of different regions, and traffic load can be balanced from upstream regions rather than only from downstream regions. Extensive simulation experiments validate the effectiveness and advantages of our proposed scheme in terms of connection cost, network robustness, load balance degree, and network longevity.

Wireless vital sign sensing (VSS) of respiration, capturing signal variations caused by chest movements, is sensitive to signal propagation paths, with current research primarily focusing on coarse-grained changes in environments. However, the frequently changing relative positions between the chest and devices result in varying signal propagation degradation, contributing to adverse monitoring states such as non-line-of-sight, long distance, blind spots, and suboptimal orientations, making the sensing performance highly susceptible. In this paper, we propose a robust VSS scheme with WiFi under varying signal degradation (RoSe). The core mechanism lies in adaptively recovering signals by learning from expert respiratory features under direct-path propagation. For broader respiratory patterns, we employ a diffusion model to expand the manually collected expert database under ideal conditions. To amplify the respiration features, we establish a signal model with multi-dimensional information, including climbing speed, evolution amplitude, peak position, and duration, providing effective guidance for subsequent recovery. The expert characteristics serve as a reference, from which imitation learning identifies strategies to adaptively recover respiration features disrupted under diverse adverse states. According to our proposed robustness metric, accounting for performance variance, balance, and loss, experimental results on 20,000 samples demonstrate that RoSe achieves superior robustness across varying signal degradation.

Indoor localization and trajectory tracking in multi-story, energy-constrained IoT environments such as smart healthcare and industrial monitoring remain challenging. This paper investigates how to achieve reliable, fine-grained localization under practical cost and power constraints that necessitate single-gateway LoRa deployments. Existing methods such as Wi-Fi, BLE, and UWB require dense infrastructure or incur high power consumption, while conventional RSSI fingerprinting lacks robustness and cross-domain generalization. Generative methods such as GANs and VAEs typically exhibit training instability and produce oversmoothed fingerprints, which limits their applicability to complex indoor environments. To address these gaps, we present D-Trace, a lightweight trajectory tracking system that employs a conditional diffusion model to generate high-fidelity RSSI fingerprints guided by spatial priors. The system introduces an RSSI + feature representation that enhances discriminability and robustness, reduces manual data collection, and enables cross-domain generalization across floors and LoRa configurations. Extensive experiments in a multi-floor building show that D-Trace achieves 95.94% localization precision with a 0.5 m mean error under sparse deployments, and maintains up to 90.3% precision with a 3.26 m average error in cross-domain scenarios. These results validate the system’s practicality, scalability, and robustness for resource-constrained IoT deployments, providing a cost-effective solution for intelligent indoor tracking.