Distributed Acoustic Sensing (DAS) systems provide rich vibration information for industrial monitoring. However, raw one-dimensional signals are often noisy and exhibit complex and heterogeneous temporal characteristics, making it difficult for traditional models to extract discriminative features and limiting classification performance and generalization. To address these challenges, we propose a DAS signal classification frame-work that integrates dynamic time-frequency representation with a learnable multi-source feature fusion network. First, a dynamic Short-Time Fourier Transform (STFT) method is introduced, which adaptively adjusts the window size and overlap ratio according to the input signal length. This design ensures consistent time-frequency resolution while avoiding distortions caused by padding or resampling, thereby enhancing spectral discriminability. Second, we design MFOFUNet, a dual-branch UNet-based architecture that separately encodes the dynamic spectrograms of intensity and phase channels and integrates them through a multi-stage fusion mechanism. This structure preserves modality-specific complementary information while enabling effective cross-modal interaction. Furthermore, a lightweight learnable fusion module is proposed to adaptively balance the contributions of different modalities at each encoding stage, achieving improved accuracy and stability with minimal additional complexity. In addition, a dual-stage data augmentation strategy is developed, combining 1D signal-level and 2D spectrogram-level enhancements to improve robustness. Extensive experiments on a challenging 10-class DAS dataset demonstrate that the proposed method achieves superior performance, reaching 89.33±4.27% accuracy, 89.02±4.50% F1-score, and 95.57±2.40% mAP under 10-fold cross-validation, outperforming existing mainstream approaches and showing strong potential for industrial sensing applications.

In the modern digital landscape, integrating geographic locations and textual descriptions within a geo-textual dataset enhances location-based services (LBS) via spatial keyword queries, as these queries combine spatial and textual information to deliver more precise and personalized results. Additionally, the advent of cloud computing allows data owners to outsource data management and services to the cloud, boosting scalability but introducing efficiency challenges due to complex encryption. Although many schemes have been proposed for spatial keyword queries on encrypted geo-textual data, none supports matching a query keyword set with the keyword sets of multiple objects, a common query type in LBS. Imagine a user seeking to rent a house close to his/her workplace, with easy access to conveniences like supermarkets. By using nearby-fit spatial keyword queries, we can match the desired house with a house-type target object and its nearby amenities, offering more practical and flexible recommendations than traditional spatial keyword queries. Hence, in this article, we introduce an efficient and privacy-preserving scheme called the privacy-preserving weighted nearby-fit spatial keyword (PWNSK) query scheme. First, we design a target-oriented spatial keyword (TOSK) tree for data organization and a TOSK tree-based weighted nearby-fit spatial keyword (WNSK) query algorithm for efficient pruning by simultaneously utilizing locations, keywords, and distances from nearby objects to target objects. For privacy, we develop several protocols, including one for polynomial coefficient re-encoding, based on polynomial coefficient encoding and fully homomorphic encryption. Building on these protocols, we introduce our PWNSK scheme. A thorough security analysis confirms its robustness, while extensive experiments also showcase its effectiveness.

With advancements in AI, smart home gyms are becoming increasingly popular for providing fitness assistance in indoor environments. In this research, we propose a layer-by-layer framework, called Wi-Fitness, which bridges video perception with Wi-Fi sensing for smart fitness. At the data preprocessing layer, the singular value decomposition-based channel state information denoising mechanism is leveraged to do the Wi-Fi data calibration. Diverse and high-quality training samples are generated by a random quantization-based data augmentation method. At the bimodal fusion layer, the heterogeneity between the Wi-Fi and video is mitigated by the local attention mechanism and the bimodal feature integration mechanism. For the video modality, the attention-based spatio-temporal graph convolutional network (AST-GCN Net) is proposed to refine spatial information. The spatio-temporal semantic alignment module is proposed to transfer spatial information from video to Wi-Fi and maintain temporal consistency across modalities. The fitness assessment layer provides exercise visualization. The generalization of Wi-Fitness is enhanced by layer-by-layer collaboration. Wi-Fitness demonstrates its effectiveness by achieving an average F1-Score of 92.68% in three typical indoor environments.

The existing commodity Wi-Fi-based human gait recognition systems mainly focus on a single subject due to the challenges of multisubject walking monitoring. To tackle the problem, we propose Wi-Diag, the first commodity Wi-Fi-based multisubject abnormal gait diagnosis system that leverages only one pair of off-the-shelf commercial Wi-Fi transceivers to separate each subject’s gait information and maintains an excellent performance when the scenario changes. It is an intelligent multisubject gait diagnosis system that can release an experienced doctor from heavy load work. Multisubject abnormal gait diagnosis is modeled as a blind source separation (BSS) issue, and multisubject walking mixed signals are efficiently separated by IC analysis (ICA) approach. This fact is verified by comprehensive theoretical derivation and experimental validation. In addition, CycleGAN is leveraged to mitigate the environmental dependency so that Wi-Diag can be robust when the scenario changes. The excellent performance of Wi-Diag is verified by extensive experiments. The average mean diagnosis accuracy with a maximum group size of four and various scenarios is 87.77%.

As the mobile Internet grows rapidly, location-based services (LBSs) are widely applied in the tourism and transportation fields. To fully mine the data collected, location service providers (LSPs) intend to offer various query services to users, which include eclipse query. The eclipse query can generalize nearest neighbor queries and skyline queries and allow users to set more rough and customizable preference ranges. In addition, with the boom of cloud computing, more and more LSPs hope to leverage the cloud to offer better query services. Given that the data could potentially contain confidential information, the data are required to be encrypted prior to outsourcing them. Therefore, eclipse queries need to be executed on the ciphertext. Although several schemes for eclipse queries have been proposed in existing works, they have little focus on privacy issues. To address this issue, we propose an efficient and privacy-preserving scheme for eclipse queries (EPEQ) in this paper. First, we develop a MinValue tree to construct an index for the dataset. Then, by utilizing the MinValue tree and a symmetric homomorphic encryption technique, we design a secure minimum value comparison protocol to obtain a skyline data and a secure undominated data acquisition protocol to obtain the data not skyline dominated by the skyline data. After that, we present our scheme. We analyze the security of the EPEQ scheme and perform experimental evaluations, demonstrating the security and efficiency of our EPEQ scheme.

With the popularity of location-based social networks (LBSNs), locations and social relationships have been considered to be important factors in point-of-interest (POI) recommendation services. The boom of cloud computing has driven data owners to outsource the LBSN data and the POI recommendation services to the cloud with powerful computing and storage capabilities. However, as the data usually contains sensitive information, it should be encrypted before being out-sourced, and consequently, the POI recommendation has to be processed over encrypted data. Although several privacy-preserving LBSNs-based POI recommendation schemes have been proposed, they are either inapplicable to the outsourcing scenario or have issues with the recommendation accuracy. Aiming at addressing these issues, in this paper, we propose an efficient and privacy-preserving LBSNs-based POI recommendation scheme (PLPR). Specifically, we first index the users' social relationships with Bloom filters and then organize the user resident location and social relationship dataset into a Vantage Point (VP) tree. Then, we design an efficient LSBNs-based POI recommendation algorithm based on the VP tree. After that, we design a privacy-preserving range determination protocol (PRD) and a privacy-preserving neighbor determination protocol (PND) to respectively protect the privacy of locations and social relationships in the designed algorithm and propose our PLPR scheme. Security analysis shows that our scheme is privacy-preserving, and performance evaluation demonstrates that our scheme is also efficient.

As a key task in autonomous driving, 3D object detection based on LiDAR-camera fusion is expected to achieve more robust results by the complementarity of the two sensors. However, LiDAR-camera fusion is non-trivial. An existing problem for this type of detector is that the scale and receptive field of LiDAR point features and image features are not matched, leading to information deficiency or redundancy in fusion. This paper proposes a Point-based Pyramid Attention Fusion (PPAF) module for LiDAR-camera fusion to solve the problem. The PPAF module learns corresponding image features of LiDAR points with a matched scale based on the image feature pyramid and attention mechanism for a better effect of fusion. Furthermore, based on the PPAF module, a new LiDAR-camera fusion-based 3D object detector named FS-Net is proposed, a two-stage detector with LiDAR voxel-based RPN and refinement network based on enriched LiDAR-camera features. Experiments on two public datasets demonstrate the effectiveness of our approach.

The Internet of Things (IoT) boom has enabled Internet Service Providers (ISPs) to collect an enormous amount of high-dimensional data. Performing range queries on such data can effectively reuse them to help ISPs offer better services. Owing to the low cost and high resource utilization of cloud computing, an increasing number of ISPs are inclined to outsource data and services to it. However, as the cloud is not fully trusted, data need to be encrypted before being outsourced, which inevitably hinders many query services, e.g., range queries. Various schemes were proposed for privacy-preserving range queries, yet they struggled to extend to high-dimensional scenarios and did not support dimension selection. Aiming at this challenge, in this article, we propose an efficient and privacy-preserving high-dimensional range query scheme (PHRQ) based on an iMinMax tree while supporting dimension selection. Specifically, we first build an iMinMax tree for high-dimensional data and utilize a symmetric homomorphic encryption technique to design a suite of privacy-preserving protocols to achieve secure high-dimensional range queries. Then, we design a sub-dimensional range determination protocol to support dimension selection. Further, based on the iMinMax tree and our privacy-preserving protocols, we propose our PHRQ scheme. Finally, security analysis shows that our scheme is privacy-preserving, and performance evaluation demonstrates that our scheme is efficient in high-dimensional range query processing.

Maintaining local consistencies can improve the efficiencies of the search algorithms solving constraint satisfaction problems (CSPs). Comparing with arc consistency which is the most widely used local consistency, stronger local consistencies can make the search space smaller while they require higher computational cost. In this paper, we make an attempt on the compromise between the pruning ability and the computational cost. A new local consistency called singleton strong bound consistency (SSBC) and its light version, light SSBC, are proposed. The search algorithm maintaining light SSBC can outperform MAC on a considerable number of problems.