Human Activity Recognition (HAR) has become increasingly important in healthcare, smart homes, and human–computer interaction applications. However, traditional vision-based approaches suffer from privacy concerns and high deployment costs, while single-sensor methods are often limited in robustness and generalization capability. To address these challenges, this study proposes a multimodal HAR framework that integrates a 2×2 Wi-Fi Channel State Information (CSI) array with wearable inertial sensors. The proposed 2×2 CSI array enables synchronized multi-channel acquisition and fusion, improving signal stability and reducing packet loss in complex indoor environments. Meanwhile, accelerometer and gyroscope data are collected from a smartwatch and combined with CSI signals to construct a comprehensive multimodal representation. A hierarchical deep learning architecture, termed M2HAR-Net, is designed to effectively extract and fuse spatial–temporal features from heterogeneous modalities, capturing complementary motion characteristics. To enhance computational efficiency and real-time responsiveness, an ablation study on PCA-based frequency-domain reduction demonstrates that retaining only the first principal component preserves discriminative information while significantly reducing inference overhead. Experimental results on a dataset collected from four participants show that the proposed method achieves an overall accuracy of 98.77% across nine daily activities. Comparative evaluations with several state-of-the-art models further validate the effectiveness and robustness of the proposed framework. These findings indicate that efficient multimodal fusion can substantially improve HAR performance, while future work will focus on cross-environment generalization and large-scale validation.

This paper introduces a comprehensive health monitoring and pedestrian localization solution that combines a foot-mounted inertial measurement unit (IMU) with a wrist-worn health monitoring device. The system uses pedestrian dead reckoning (PDR) along with an advanced gait recognition algorithm to deliver continuous 3D localization, achieving an accuracy of less than 1 meter in both indoor and outdoor settings. The wrist-worn sensor integrates electrocardiography (ECG), photoplethysmography (PPG), and an accelerometer to monitor vital signs in real time and detect falls. This integrated approach provides a reliable solution for healthcare monitoring, location tracking, and emergency response applications.

Wearable sensor–based human activity recognition (HAR) is essential for navigation and industrial automation, where real-time data processing is critical. However, the inherent nonlinearity and noise in sensor data make accurate and fast activity detection a challenging task, and traditional feature extraction techniques often struggle to capture the underlying dynamic changes. In this paper, we introduce WavI2I, a novel framework that enhances HAR by first applying the Continuous Wavelet Transform (CWT) to convert inertial sensor signals into time-frequency planes, which are then processed by Convolutional Neural Networks (CNNs). To further optimize feature extraction, we incorporate a nonlinear scale generator, ensuring a balanced focus on both high- and low-frequency components. These innovations strengthen the CNN’s ability to identify critical features, thereby improving recognition accuracy. Experiments on the UCI-HAR dataset confirm that WavI2I surpasses existing methods in terms of accuracy, computational efficiency, and model simplicity.

Pedestrian Dead Reckoning (PDR) technology demonstrates significant application value in smart city location services and IoT terminal positioning due to its signal-independent operation, autonomous navigation capability, and anti-interference advantages. However, existing shoulder-mounted inertial measurement units (IMUs) encounter gait characteristic modeling errors during practical deployment, particularly manifesting as nonlinear error accumulation caused by limited step-length prediction accuracy. To address this technical challenge, this study proposes a step-length estimation model based on residual neural networks (ResNet) with limited-sample training. The architecture achieves precise step-length prediction across various motion states through temporal feature extraction and multi-rate motion pattern analysis. Experimental results validated by five independent test sets demonstrate that the system achieves a relative displacement estimation error below 0.6%, with the mean absolute error (MAE) of single-step length prediction consistently remaining under 0.045 meters. Analytical verification confirms that the proposed step-length estimation method significantly enhances the step-length measurement accuracy of shoulder-mounted IMUs, providing an effective technical solution for high-precision indoor positioning of IoT devices.

This paper presents a practical TDOA-based method for UWB anchor localization, aiming to simplify the process of determining anchor positions, reduce costs, and improve efficiency. By utilizing a small number of tag position coordinates and the TDOA information between anchors and tags, and by introducing a weighted least squares approach, this method can quickly and effectively solve for UWB anchor coordinates even without any prior knowledge of their initial positions. Experimental results demonstrate that the proposed method can achieve positioning accuracy within 1 meter, with the rectangular trajectory demonstrating the highest stability and accuracy, as indicated by the lowest RMSE (Root Mean Square Error) and HDOP (Horizontal Dilution of Precision) values. Future research will focus on optimizing the algorithm under complex environmental conditions, integrating data from multiple sensors such as LiDAR or cameras, enhancing real-time performance, and developing user-friendly interfaces. These efforts aim to further enhance the method's robustness and practical applicability. Additionally, comparative experiments in diverse scenarios will be conducted to validate the effectiveness and applicability of the proposed approach.

While smartphone-based fingerprinting techniques have emerged as promising solutions for indoor localization, their efficacy remains constrained by suboptimal fingerprint database quality and algorithmic limitations in 2D coordinate estimation. Conventional approaches suffer from laborious data collection processes, constrained accuracy, and diminished reliability over extended periods. To address these challenges, this study proposes a novel attention-enhanced LSTM architecture synergistically integrating heterogeneous sensor data (WiFi, barometric pressure, and magnetometer) to achieve simultaneous planar localization and multi-floor identification. A dedicated foot-mounted inertial measurement system is introduced to streamline fingerprint database construction by enabling efficient sparse data acquisition through collaborative smartphone-device interactions. The developed LSTM-Attention framework demonstrates superior performance in initial position matching precision, accelerated model convergence, and enhanced trajectory consistency through adaptive feature weighting. Comprehensive evaluations across multi-story academic and office environments reveal pedestrian localization accuracy within 1.5 meters (horizontal) and floor discrimination success rates exceeding 95%, thereby advancing the state-of-the-art in smartphone-based 3D indoor positioning systems.

Over-the-air computation (AirComp) leverages the inherent superposition property of wireless multiple-access channels to enable direct signal aggregation from massive users. However, unfavorable channel conditions can severely degrade the computed mean square error (CMSE). To address this limitation, we introduce a reconfigurable intelligent surface (RIS) into the AirComp system. Unlike prior works that assume full signal reflection by each RIS element (RE) regardless of its phase shift, we adopt a practical model accounting for the coupling between the amplitude and phase of each RE. Based on this model, we formulate an optimization problem to minimize the CMSE by jointly optimizing transceiver design and the RIS reflection matrix. To tackle this highly nonconvex problem, we propose a dual-loop optimization framework, where the outer loop employs the genetic algorithm to obtain a near-optimal RIS reflection matrix and the inner loop uses an alternating optimization approach for transceiver design. Simulation results demonstrate that the proposed dual-loop algorithm outperforms baseline methods in reducing the CMSE.

Substations are critical infrastructures in modern power systems, requiring accurate and reliable personnel positioning to ensure operational safety and inspection efficiency. Pedestrian dead reckoning (PDR) technology has emerged as a preferred solution for real-time seamless positioning due to its autonomous, anti-jamming and passive characteristics. However, in practical applications, the PDR algorithm faces challenges such as the inability to provide absolute positioning due to unknown starting points and cumulative errors over time. To address these challenges, this paper proposes a head-mounted pedestrian positioning system (HPPS) integrating fisheye camera/real-time kinematic (RTK)/PDR based on factor graph optimization (FGO). First, a skyward-facing fisheye camera identifies non-line-of-sight (NLOS) signals caused by architectural obstructions, improving the absolute positioning accuracy of RTK. Next, an adaptive pedestrian gait detection and step length estimation algorithm based on head-mounted inertial measurement units (IMU) is developed to enhance PDR robustness. Finally, the FGO framework integrates the inputs of the fisheye camera, RTK, and PDR to achieve seamless indoor-outdoor positioning. Experimental results demonstrate that the proposed system achieves horizontal root mean square error (RMSE) values of less than 0.42 m and 0.35 m in two distinct outdoor substation environments and less than 1.36 m in indoor-outdoor scenarios. This work highlights the integration of precise positioning technologies within the Internet of Things (IoT) framework, advancing smart grid applications and enabling effective location tracking in complex environments.

Fingerprint localization is a critical method for indoor positioning that has garnered considerable attention. Traditional methods for constructing fingerprint databases and matching algorithms frequently exhibit inefficiencies and limitations, which can undermine both the accuracy and the robustness of localization systems. This paper introduces an innovative indoor pervasive localization method leveraging deep learning. We employ a hybrid system of foot-mounted positioning devices and smartphones to efficiently create a universal fingerprint database, subsequently utilizing LSTM-based deep learning methods for accurate pedestrian location matching. Experiments conducted within a standard academic building demonstrate that our proposed method can more accurately map the indoor movement trajectories of pedestrians. The localization results indicate a horizontal accuracy of 2.5 meters and a vertical accuracy of 0.2 meters. Notably, our method shows a 10% improvement in horizontal accuracy and an 18% improvement in vertical accuracy over WiFi-only approaches. Furthermore, when compared to the classical Random Forest models, our method achieves performance enhancements of 20% in horizontal and 15% in vertical accuracy.

Foot-mounted inertial pedestrian positioning system (FIPPS) is increasingly important in the Internet of Things (IoT) and smart cities because of the passive, anti-interference, and fully automatic advantages. However, FIPPS faces the problem of unknown initial positions and accumulative errors in practical applications, which makes the positioning trajectory inaccurate and unable to be matched in a well-defined coordinate system. A seamless indoor–outdoor inertial pedestrian positioning method based on smartphone precise point positioning (PPP)/3-D map/barometer augmentation is proposed to solve this problem. The main contributions include the following: 1) a 3-D-mapping-aided PPP (3DMA PPP) method is proposed to detect and eliminate non-line-of-sight (NLOS) signals to provide high-precision initial coordinates for FIPPS, which enables positioning trajectories to be matched in the unified WGS-84 coordinate system; 2) the adaptive zero-velocity update (ZUPT) approach based on neuro-fuzzy inference is used to precisely identify the stance phases of various gait patterns to suppress position and velocity error accumulation; 3) a barometer and 3-D building model-based height constraint algorithm is applied to further improve the height estimation of the FIPPS; and 4) an Android application named FYTECH is developed for data transmission between the smartphone and FIPPS and supports the online display of pedestrian positioning trajectories. In a seamless outdoor–indoor experiment with multiple motion patterns, including walking, running, elevators and stairs, we demonstrate that the proposed method can achieve RMS errors better than 1.17 and 1.32 m in the horizontal and vertical directions, respectively. The performance indicates that our method can conveniently fuse multisource sensor information, such as foot wearables, smartphones, and 3-D maps to greatly enhance pedestrians’ seamless indoor–outdoor navigation experience.

With the explosive development of pervasive computing and the Internet of Things (IoT), indoor positioning and navigation have attracted immense attention over recent years. Pedestrian dead reckoning (PDR) is a potential autonomous localization technology that obtains position estimation employing built-in sensors. However, most existing PDR methods assume that the smartphone is held horizontally and points to the walking direction. To solve reckoning errors caused by inconsistency of headings between walking heading and pointing of smartphone, we design an accurate and robust PDR method based on walking patterns, which is identified by multihead convolutional neural networks. In addition to adaptively adjust the threshold of step detection and select the most suitable step length model according to the results of walking pattern recognition, a novel heading estimation approach independent of device orientation is proposed. To mitigate accumulative errors, we proposed an online trajectory calibration method based on forward and backward magnetic fingerprint trajectory matching. We conduct extensive and well-designed experiments in typical scenarios, and the experimental results indicate that the 75th percentile localization accuracy of the three scenarios is 1.06, 1.08, and 1.22 m, respectively, using the commercial smartphone embedded sensor without any dedicated infrastructures or training data. Despite the intricate pedestrian locomotion, the proposed PDR method has great potential in pedestrian positioning.