Bit-Depth Enhancement (BDE) is designed to restore High-Bit-Depth (HBD) images from Low-Bit-Depth (LBD) input, but existing methods mostly fail to exploit their algorithmic advantages in extreme low-bit cases. In this paper, a new framework that combines noise shaping, latent space modeling, and adaptive weighting block (AWB) is proposed to solve the above problems. Firstly, based on the noise shaping method, Sigma-Delta (Σ-Δ) quantization is introduced to achieve low redundancy and high reversibility low-bit image construction. Second, based on the visual averaging property and Inverse Problem Transform (IPT), the conditional posterior distribution pz(HBD|LR) is introduced for the first time to model the relationship between low-resolution (LR) and HBD images in the latent space to achieve the recovery of high-frequency details. To realize effective interaction and adaptive fusion of spatial and bit information, the hierarchical feature discovery module (HFDM) is introduced in network construction to generate multiscale LBD-LR image pairs, while AWB dynamically fuses according to the amount of feature information. This framework reaches the leading level in both quantitative metrics (PSNR/SSIM) and visual quality, and provides a new idea for high-bit image generation.

Accurate modeling of proximity relationships between individuals and groups in urban spaces is critical for public safety alerts, emergency evacuation planning, commercial flow analysis, and urban planning optimization. Existing indoor localization techniques-including TOA/TDOA/AoA geometry-based methods, visual SLAM, inertial navigation (PDR), and RSSI/UWB fingerprinting-achieve high accuracy in static environments but rely heavily on large-scale manual labeling, strict clock synchronization, or dense anchor deployment, and struggle to accommodate both trajectory continuity and dynamic environmental changes. To address these limitations, we propose a dual-modal deep learning framework based on Graph Convolutional Networks and Multidimensional Scaling embedding. First, channel feature similarities are computed from CIR sequences sampled continuously by mobile terminals, and an attentionenhanced proximity graph is constructed. Then, the graph adjacency matrix is fed into the MDS algorithm to recover the relative low-dimensional layout of observation points, enabling robust reconstruction of crowd density and flow topology. To mitigate class imbalance and manage positive and negative pair ratios both within and across trajectories, we introduce a spatiotemporal pair generation strategy and employ a weighted binary crossentropy loss during training. Simulation results demonstrate that under sparse sampling and dynamic conditions, our method significantly outperforms traditional thresholding and existing unsupervised baselines in both precision and recall, and approaches the performance of supervised methods under moderate label noise. This framework offers an efficient, annotation-free solution for indoor crowd density estimation and flow analysis.

Today, Wi-Fi positioning has been widely deployed in many indoor localization scenarios. However, the accuracy of a specific Wi-Fi based positioning system is significantly affected by environmental factors, particularly the characteristics of the wireless communication channel. In this paper, we employ the Cramér-Rao Lower Bound (CRLB) to theoretically evaluate the accuracy of a map-information-assisted Wi-Fi positioning system. A signal propagation model incorporating wall-induced attenuation is proposed, where the non-differentiable step behavior of received signal strength (RSS) is approximated using a smooth sigmoid function. This allows the impact of walls to be analytically embedded in the CRLB formulation. Both simulation and real-world experiments are conducted to assess the proposed model. Results demonstrate that, under identical conditions, positioning accuracy near walls consistently outperforms that in wall-free regions.

In addressing the dataset requirements for indoor inertial navigation technology based on deep learning, this paper proposes a comprehensive dataset construction and data preprocessing scheme which includes equipment preparation, data collection and data alignment. To meet the demands of deep learning for data size and distribution diversity, a detailed data collection strategy is introduced. For the acquired dataset, a robust data processing methodology is employed, involving time calibration, data alignment and coordinate system transformation to synchronize sensor data with groundtruth data. The dataset is then trained and tested on the Baseline network, followed by ablation experiments. The results show that the dataset constructed by the proposed method has the best trajectory reconstruction performance, while the small-size dataset that does not contain complex motion patterns has obvious drift and the dataset that does not use the proposed data alignment algorithm has a very poor performance. These outcomes validate the proposed scheme’s capability to enhance the performance of indoor inertial navigation models.

GNSS has inherent shorting comings such as susceptibility to spoofing and unreliability in complex environments (e.g., city canyon) due to satellites visibility issues. With the development of the fifth generation (5G) cellular network, the capability and accuracy of cellular-based positioning has been significantly improved, providing a promising alternative to GNSS for pervasive positioning services. The exact 3-dimensional (3D) coordinates of cellular base stations are crucial for time-offly (ToF) based multilateration or angular based positioning. However, as cellular operators are unwilling to publicize BS configuration, cellular positioning users often struggle to obtain precise BS coordinates. In this paper, we propose a multipath-assisted BS calibration scheme based on the Probability Hypothesis Density (PHD) filter. By utilizing ToA signals, we estimate the spatial locations of 5 G BSs and synchronization errors in the environment. In addition, virtual anchors (VA) are exploited in our model as additional prior information to further enhance positioning accuracy in fusion positioning. The positioning performances of the proposed method are evaluated based on 5 G ToA signals. Simulation results indicate that when the Optimal Subpattern Assignment (OSPA) metric falls below a certain threshold, synchronization errors are less than $5 \%$. The effectiveness and accuracy of the proposed scheme for positioning in urban environments are also analyzed and discussed.

As e-commerce continues to grow, comprehending consumer emotions has be-come crucial, directly influencing product sales and corporate reputation. Yet, the high dimensionality and complexity of e-commerce platform comment data pose significant challenges to conventional sentiment analysis techniques. In response, this paper presents an innovative sentiment analysis model for e-commerce re-views, leveraging GPT-3.5 embeddings, convolutional neural networks (CNN), bidirectional long short-term memory (BiLSTM) networks, and Self-Attention mechanisms. Initially, GPT-3.5 is employed to transform input text into dynamic word vectors, thereby capturing subtle semantic distinctions. Subsequently, the CNN layer is utilized to extract textual emotional features, while the Self-Attention mechanism enhances the model's sensitivity to subtle emotional shifts. The BiLSTM layer further captures bidirectional, backward and forward contex-tual information, offering a holistic grasp of textual sentiment. Emotional features are ultimately classified using the Softmax function. Relative to existing models, our proposed model exhibits superior performance in e-commerce dataset exper-iments, demonstrating notable enhancements in accuracy, precision, recall, and F1 scores. This validation underscores our solution's effectiveness in conducting sentiment analysis within the e-commerce domain.

Multipath propagation greatly affects the accuracy of time of arrival (ToA)-based indoor positioning when line-of-sight (LOS) signals are only used. In this paper, we present a novel real-time and low computation complexity multipath-assisted ToA positioning method, namely MSC-MAP. The delays of reflected signals are taken as additional spatial observations to compensate for an insufficient number of physical transmitters to locate a moving user equipment (UE). Virtual anchors are used to model the propagation path of reflected signals, whose locations are obtained via a multi-state constraint estimator, along with the trajectory of UE. In addition, we demonstrate the mismatch problem in data association and its impact on positioning performance. To achieve real-time processing, we propose two robust multipath-assisted positioning methods with mismatch alleviation by randomly selecting subset and constraint relaxation respectively, to meet various computational complexity requirements. Simulation results show that, for the MSC-MAP method, the mean square error of the position is generally less than 0.2 m in challenging indoor environments. Among mismatch alleviation algorithms, positioning error is reduced by 69% even when the percentage of mismatched measurement data is as high as 42%. The proposed algorithms can also efficiently handle signals with non-Gaussian impairments, a common characteristic in real-world data. Moreover, these algorithms can substantially improve positioning performance while adding minimal computation time in the presence of measurement mismatches, outperforming state-of-the-art methods utilizing different data association techniques.

The complex topography of urban canyons with many reflectors and scatterers makes it challenging to detect low-altitude, smaller, and slow-speed targets. In this paper, we present a novel multipath-assisted passive detection scheme based on the global navigation satellite system signals in urban canyons. We first propose an information-level target detection scheme, where a binary hypothesis test is conducted according to variation in the received signal given the presence or absence of targets in the environment. To take usage of multipath components (MPCs) in the proposed scheme, we introduce virtual anchors to model reflected signals’ propagation paths. We also introduce the reconfigurable intelligent surface to artificially improve the reflective environment and enhance the quality of received MPCs. The detection performance indicators are analyzed theoretically. Simulation results show that the proposed schemes respectively reach 90% and 94% detection probability at a signal-to-noise ratio of 5 dB. The RIS-based method outperforms the multipath-assisted method when the RIS error is less than 0.41 m.

In healthcare facilities, indoor localization technology has a broad range of applications. Traditional Pedestrian Dead Reckoning (PDR) and WiFi fingerprint-based methods each have their limitations. To address these challenges, this study introduces a multi-source fusion indoor localization system that uses a Factor Graph to integrate inertial positioning algorithms with WiFi fingerprint-based localization. The system processes accelerometer and gyroscope data using a data-driven PDR algorithm. For WiFi localization, considering that the extensive data collection required is a significant barrier to the deployment of WiFi-based localization methods, the proposed approach applies Gaussian process regression techniques to limited WiFi fingerprint data, significantly reducing initial deployment costs and enhancing accuracy. Finally, the entire system employs a Factor Graph for the integration of the data-driven PDR and WiFi fingerprint localization results. Experimental results show that, compared to using only inertial or WiFi data for localization, this method significantly improves localization accuracy. The findings suggest that this approach could prompt the utilization of indoor localization technology in healthcare facilities.

Wi-Fi fingerprint-based positioning is widely used caused by its low cost and ease of implementation when Global Positioning System technology struggles to obtain signal and accuracy in indoor environments. A prerequisite for Wi-Fi fingerprint-based positioning is the availability of an accurate fingerprint database. However, in practice, access to a large proportion of the area is restricted, leading to a rapid degradation of the positioning accuracy of existing interpolation techniques in the absence of reference data in the vicinity of the area. In this paper, we propose an enhanced propagation model constrained fingerprint patching algorithm for incomplete RSS fingerprint databases, aiming to extend the coverage of fingerprints and reduce the overall workload of survey. The algorithm integrates a map-assisted wireless signal propagation models with incomplete spatially sampled fingerprint data to construct a propagation model for each Wi-Fi access point (AP). Recalculate the received signal strength at each reference point for each AP in the fingerprint database according to the specific model. The fingerprint information for the unrecorded regions is then refined by interpolation and the fingerprints initially collected are optimized. We experimentally validate the effectiveness of the proposed interpolation algorithm and compare it with the conventional Support Vector Regression (SVR), Gaussian Process Regression (GPR), and rational quadratic kernel (RQK) methods. The results show that the positioning performance of proposed method is improved by 18.1%, 23.1% and 39.7% in the blank area relative to RQK, GPR and SVR, respectively. In addition, our method outperforms other algorithms with stable performance in different scenarios, especially in scenarios containing obstacle walls, and the performance benefit appears more notable in the complicated indoor scenario with cramped rooms.

Multipath-assisted positioning is a promising way to realize robust and accurate indoor positioning, taking advantage of the environmental information carried by multipath signals. In this paper, we propose a novel multipath-assisted time-of-arrival (TOA) positioning method for complex indoor scenarios without a-priori knowledge of the floor plan. We use virtual anchors (VAs) to model the propagation path of reflected signals. The trajectory of the user equipment (UE) and the locations of VAs are iteratively estimated using two unscented Kalman filters (UKFs), considering the changeable visibility of VAs due to the birth and death of multipath components (MPCs). To use TOA measurements of MPCs as input for the update phase of the UKF, we present a data association method based on multipath peak flow tracking by baseband signal processing to establish the correspondence between measurements and VAs. We derive the analytical solution using the received power of MPCs for multipath tracking, which can realize low computational complexity data association. Simulation results show that, most of the MPCs can be correctly tracked using the peak flow method, even if some MPCs are densely distributed. For the proposed iterative UKF, the mean square error of the UE's position, with associated measurements obtained by the peak flow method as input, is generally less than 0.42 m in the complex indoor scenario.

In recent years, the Visual Inertial Odometry (VIO) technology has attracted attention as a support technology that improves medical experience and management efficiency. However, the performance of most VIO systems will drop drastically when the light intensity changes significantly or there are few texture features from the images. This paper designs a visual-inertial fusion-based navigation Indoor Positioning system to deal with the challenging scenario. It loosely coupled an inertial sensor-based pedestrian dead reckoning (PDR) model with the VIO model via a robust particle filter. The state estimation of the particle filter is based on the PDR model. The VIO model is used for the measurements of the particle filter. It compensates the gross errors of the VIO with a visual error propagation model which is established according to the posterior observation residuals of visual feature points. It is verified through experiments that the PDR/VIO fusion indoor positioning system based on the robust particle filter implemented in this paper has improved positioning accuracy and strong ability to deal with complex scenes.

On-demand streaming of high-quality video content is a widely anticipated vehicular infotainment service. How to reduce the cost of content streaming is a primary concern of the service providers, but is still an underinvestigated subject in the literature. This paper proposes an integrated mobile streaming and caching scheme that jointly leverages two communication service tiers and on-board caching resource for cost reduction. Algorithms are presented to achieve optimal buffering at the session level and optimal caching at the device level. An analytical framework is established to characterize the average cost as a function of the streaming rate in a large scale network. Numerical results demonstrate how the “cost-streaming rate” function changes with vehicle density, network congestion level, content length, and average packet transmission time. We learn an important insight that there is a minimum cost threshold even when the streaming rate approaches zero. We also show that the proposed protocol can effectively reduce the overall cost when the network is not congested. Our findings can provide useful guidelines for the business planning and operation of vehicular content streaming services.

Aimed at the problem of complex progress for indoor map building, this paper proposes an RSS-based indoor map construction algorithm only using information of WiFi fingerprint. The indoor map construction problem is then transformed into a classification problem of reference points in fingerprint database. With the aid of hierarchical classification system consisting of single-AP-based base classifier and multi-AP-based combination classifier, the accuracy of the classification results is guaranteed. Moreover, an accurate line segment feature map is obtained by identifying demarcation line between two classes from the hierarchical classification system. The simulation experiments are used to evaluate the effectiveness of the classification algorithm and the feasibility of the map construction algorithm.

Over the past decade, deep learning developed rapidly and had significant impact on a variety of application domains. It has been applied to the field of human activity recognition to substitute for well-established analysis techniques that rely on handcrafted feature extraction and classification methods in recent years. However, less attentions have been paid to the influence of training data on recognition accuracy. In this paper, we assessed the influence factors of data volume and data balance in human activity recognition when using deep learning approaches. We evaluated the relationship between data volumes of training dataset and predict accuracy of deep learning algorithms. Given the impact of the data balance between activity categories on the recognition accuracy, we modified the SMOTE algorithm so that it can be applied to human activity recognition. Results show that when the data volume is small (<;4M), the recognition accuracy increased quickly with the increase of the quantity of training data. However, the growth trend of recognition accuracy slows down when the data quantity reaches 4 million. Further increase the data volume does not significantly improve the activity recognition performance. So we can conclude that 4 million data volume can ensure a sufficient accuracy for human activity recognition. Meanwhile, the data set balance operation can not only improve the recognition accuracy of minority categories, but also helps to increase the overall accuracy.

High accuracy in indoor navigation with foot-mounted sensors attracts a lot of researchers in the last decades. Most indoor positioning schemes based on strap-down inertial navigation can only be used for normal walking. This paper present a 3D foot-mounted inertial navigation system, which can meet the challenge of the multi-gaits. During walking, the foot will have a contact with the ground in every step, in which time, the velocity of foot is zero. The correctness of zero velocity detection is important for drift removing in pedestrian dead-reckoning based inertial pedestrian indoor position systems. Previous algorithm of zero velocity detection is hard to handle the gaits variety. In this paper, by analyzing the inertial data from different modes of motion, a heuristic zero-velocity detection algorithm is designed. The algorithm can accurately detect the zero-velocity time of pedestrians among a variety of gaits. Then the speed and the displacement are updated in the Kalman Filter. Moreover, the barometer is fused with accelerometer for the calculation of height and achievement the 3D trajectory tracking. The experimental results show that the average distance error is 2.59%, the average distance error is 5.78% during running and the average height error is about 0.2m when the pedestrian is going stairs.

A smart-phone based hand-held indoor positioning system is presented in this paper. The system collects data using the accelerometers, gyroscopes, barometers and gravity sensors embedded in the smart-phone. The accelerometer and gravity data are used for zero-velocity detection and calculating the vertical displacement of each walking step, and then the inverted pendulum model is applied to calculate the step length of every step. The angle of direction is estimated by processing gyroscope data with the quaternion method. The step length and the direction angle of each step are combined to determine the coordinates of each step. The barometer is used for measuring the height information. A Kalman filter is used in zero-velocity-update (ZUPT) to reduce the vertical speed offset caused by accelerometer drift errors. Wifi is also fused in our system. In order to guarantee the accuracy, map information and magnetic field information are used in the navigation systems. The experiment results show that we obtained high precision results with common hand-held smart-phone often seen on market.

Frequency estimation of a single tone signal is important in many applications. In this study, a frequency estimation algorithm is proposed for single tone signals modulated with unknown data. The proposed unbiased frequency estimator (UFE) uses discrete Fourier transform (DFT) coefficients to calculate the accurate frequency offset. To achieve better performance in a noisy environment, a biased frequency estimator (BFE) is obtained by applying trigonometric approximation to UFE. The Taylor expansion of BFE is also proposed to meet the computational complexity requirement in some real‐time applications. The performance of the proposed algorithm is evaluated through Monte Carlo simulations, compared with conventional interpolated DFT‐based algorithm and typical non‐data‐aided frequency estimation algorithm.