6G-driven Internet of Things (IoT) networks demonstrate enormous development potential. However, complex and diverse IoT devices also pose severe cybersecurity challenges. To address the problem of low reliability and trustworthiness of existing intrusion detection systems (IDS) in large-scale, low-latency scenarios of 6G-IoT networks, this paper proposes a local-global hierarchical Framework for intrusion detection. First, a reliable feature processing algorithm is designed based on fusion features and a multi-objective collaborative optimization mechanism oriented toward normal behavior modeling, enabling the binary classification model to effectively detect unknown types of attack traffic even under data anomaly conditions. Second, an adaptive classification trustworthiness assessment method based on classification information entropy and collaborative verification is proposed, which can accurately locate and upload suspicious traffic during the classification process. Finally, we propose a gradient-oriented feature decoupling module that effectively improves the feature learning capacity of network attack detection models in global servers for minority class samples. Experimental results demonstrate that our proposed method achieves optimal performance on five processed IDS benchmark datasets, with local IoT devices achieving an F1 score of 98.5% and global servers achieving an arithmetic mean F1 score of 96.6%, providing strong support for the reliable operation of IDS in 6G-IoT networks.

Federated learning (FL) enables edge devices to collaboratively train a global model without sharing raw data. In the edge environment, resource constraints hinder efficient training and aggregation of FL. Although prior studies have established model pruning as a practical strategy to reduce resource demand, the parameter symmetry problem (i.e., permutation-equivalent parameter orderings in neural networks) remains underexplored. Without addressing this symmetry, pruning on heterogeneous clients results in aggregation mismatches and degraded accuracy. In this paper, we propose Attention-guided One-shot Pruning for Federated Learning (FedAOP) to address these challenges. First, we design an attention module that integrates spatial and channel attention to highlight critical spatial responses and evaluate channel importance. Then, leveraging these importance scores, we propose an attentive pruning algorithm to generate client-specific models, thereby reducing resource consumption. Furthermore, we introduce an aggregation algorithm with attention matching, thereby mitigating the adverse effects of parameter symmetry under heterogeneous pruning. We implementFedAOPon a benchmark FL platform. The experimental results on benchmark datasets demonstrate thatFedAOPconsistently outperforms state-of-the-art baselines by up to 11.3% in accuracy while reducing the average model footprint by 32%.

Feature selection plays a crucial role in data mining and machine learning. However, evident challenges exist: (1) current methods cannot autonomously identify the optimal feature set, requiring manual parameter adjustment based on the learning algorithm; (2) heuristic methods, which are widely used, often struggle to ensure the maximization of the objective function. To address these challenges, this paper proposes a density cluster-based feature selection (DCFS) method leveraging information theory, which involves the application of artificial intelligence (AI) in the clustering process. First, a novel initial feature selection method that maximizes feature-relevance and feature-difference is introduced to automate the selection of an initial feature subset. Second, a new density-centric automatic clustering (DAC) algorithm, an AI-based clustering approach, is proposed. This algorithm synthesizes non-parametric density estimation, decision graph-based density center selection strategies, and adaptive domain search clustering methods to enhance the precision and robustness of clustering outcomes. Third, a feature space selection method based on maximizing feature relevance is established to construct a comprehensive feature subspace. This feature space selection method converts the maximization of the objective function into an automated density clustering process, facilitating the automatic selection of the most optimal features. Extensive experiments conducted across 14 datasets have demonstrated the superior performance of the proposed DCFS in terms of effectiveness and robustness. To the best of our knowledge, this paper is the first work attempting automatic feature selection through clustering, thus pushing the frontier of feature selection algorithm development.

In this study, we address the challenge of traversability analysis for autonomous vehicles in diverse environments, leveraging LiDAR sensors. We propose the Transformer-Voxel-Bird’s eye view (BEV) Network (TVBNet), a novel dual-branch framework designed to increase the accuracy and versatility of such analyses in both urban and off-road conditions. TVBNet first preprocesses raw point cloud data through voxelization and the generation of a BEV. It incorporates a Transformer network with a rotational attention mechanism to aggregate features from multiple point cloud frames, capturing long-range correlations both within and between point clouds. Additionally, a Swin Transformer extracts the relative positional relationships in the BEV projection, facilitating a comprehensive understanding of the scene. The fusion of data from both branches via a multisource feature fusion module, which employs a context aggregation mechanism based on a residual structure, allows for robust local to global contextual understanding. This approach not only improves the extraction of correlation features between 2D BEV and 3D voxel data but also demonstrates superior performance on the challenging off-road dataset RELLIS-3D and the urban dataset SemanticKITTI.

Automatically detecting human mental workload to prevent mental diseases is highly important. With the development of information technology, remote detection of mental workload is expected. The development of artificial intelligence and Internet of Things technology will also enable the identification of mental workload remotely based on human physiological signals. In this article, a method based on the spatial and time-frequency domains of electroencephalography (EEG) signals is proposed to improve the classification accuracy of mental workload. Moreover, a hybrid deep learning model is presented. First, the spatial domain features of different brain regions are proposed. Simultaneously, EEG time-frequency domain information is obtained based on wavelet transform. The spatial and time-frequency domain features are input into two types of deep learning models for mental workload classification. To validate the performance of the proposed method, the Simultaneous Task EEG Workload public database is used. Compared with the existing methods, the proposed approach shows higher classification accuracy. It provides a novel means of assessing mental workload.

Beam pumping units (BPUs) are key equipment in oilfield production. Currently, many fault diagnosis methods for BPUs have been developed, and most of them are based on feature or image classification of indicator diagrams. However, low-quality monitoring data and the limited proportion of effective pixels in indicator diagram greatly restrict the performances of these methods. This article proposes an efficient two-step fault diagnosis method for BPUs. In the first step, to overcome the impact of low-quality monitoring data, a dynamic time warping-based matching method is proposed to extract the period of the data, and then a physical model driven method optimized by Bayesian gradient descent is proposed to reconstruct the data. In the second step, to overcome the impact of the limited proportion of effective pixels in indicator diagram, a parallel deep network is proposed which directly takes the time series of the displacement and the load of BPUs as the inputs. Extensive experiments on dataset from 45 real oil wells have shown that, the proposed method can achieve the best performance compared with the state-of-the-art methods, meanwhile the computational load is only 5% of other deep learning-based methods.

Accurately estimating the state of equipment plays an important role in ensuring the efficient operation of Industrial 4.0 systems. This article focuses on monitoring the operating state and detecting the faults of beam pumping units under the condition of heavy noise within the Industrial Internet of Things. On the one hand, the equipment operating state monitoring system designed in this article uses an acceleration sensor, the signal of which contains considerable noise that greatly reduces the motion state estimation accuracy. On the other hand, the complexity of the indicator diagrams of beam pumping units makes it difficult to extract features, which limits the ability to improve the fault detection accuracy. To overcome these issues, first, a period estimation method based on self-checking that employs acceleration data is proposed to effectively overcome the influence of complex noise on the estimated data period; second, a denoising method based on a physical model is proposed to effectively reduce the influence of complex noise on the acceleration-based displacement estimation; and third, a method for detecting the faults of beam pumping units based on edge intelligence is proposed to effectively improve the fault detection accuracy while maintaining a low computational demand. Extensive experiments on real data verify the effectiveness of the proposed method. To the best of our knowledge, this is the first work to discuss the impact of the quality of data on the performance of fault detection of beam pump units.

In the area of medical image segmentation, the spatial information can be further used to enhance the image segmentation performance. And the 3D convolution is mainly used to better utilize the spatial information. However, how to better utilize the spatial information in the 2D convolution is still a challenging task. In this paper, we propose an image segmentation network based on reinforcement learning (RLSegNet), which can translate the image segmentation process into a serial of decision-making problem. The proposed RLSegNet is a U-shaped network, which is composed of three components: the feature extraction network, the Mask Prediction Network (MPNet), and the up-sampling network with the cascade attention module. The deep semantic feature in the image is first extracted by adopting the feature extraction network. Then, the Mask Prediction Network (MPNet) is proposed to generate the prediction mask for the current frame based on the prior knowledge (segmentation result). And the proposed cascade attention module is mainly used to generate the weighted feature mask so that the up-sampling network pays more attention to the interesting region. Specifically, the state, action and reward used in the reinforcement learning are redesigned in the proposed RLSegNet to translate the segmentation process as the decision-making process, which performs as the reinforcement learning to realize the brain tumor segmentation. Extensive experiments are conducted on the BRATS 2015 dataset to evaluate the proposed RLSegNet. The experimental results demonstrate that the proposed method can achieve a better segmentation performance, in comparison with other state-of-the-art methods.

Blockchain, which originated with the Bitcoin system, has drawn intense attention because of its decentralization, persistence, and anonymity. The execution environment of blockchain is isolated from the external world and requires "blockchain oracles": agents that fetch information from the out-side world. However, there is always the risk of oracles providing corrupt, malicious, or inaccurate data. To overcome this issue, this paper analyses the existing oracle working patterns, then introduces a decentralized autonomous oracle network (DAON) and its consensus protocol and noninteractive reputation maintenance and payment scheme. Meanwhile, the reputation and security monitoring services of DAON are designed to ensure the DAON provide trustworthy data services in a complex byzantine environment. The proposed method is verified by experiments on three applications. Using the proposed network, smart contracts on future blockchains could have reliable, tamper-proof inputs and outputs.

Service uncertainty modeling is an important problem of manufacturing service composition optimization, this article proposes a cloud manufacturing service composition optimization framework based on cloud-edge collaboration considering manufacturing service uncertainty. In the proposed framework, on the edge side, a model parameters estimation method of the manufacturing services' uncertainty is proposed based on Gaussian mixture regression; while on the cloud side, an intelligent evolutionary algorithm is adopted to effectively optimize the manufacturing service composition. Since the Gaussian mixture distribution is used to approximate the service availability distribution, the service uncertainty can be modeled adaptively. Compared with the previous optimization methods of manufacturing service composition with uncertainty based on the deterministic parameter models, the method proposed in this article can model the uncertainty of service more effectively, thus obtain better service composition solutions. Extensive experimental results prove the effectiveness of the algorithm.

Automatically predicting cardiovascular and cerebrovascular events (CCEs) is a key technology that can prevent deaths and disabilities. Herein, we propose predicting CCE occurrences based on heart rate variability (HRV) analysis and a deep belief network (DBN). The proposed prediction algorithm uses eight novel HRV signal features, which are calculated based on the following steps. First, the instantaneous amplitude (IA), instantaneous frequency (IF), and instantaneous phase (IP) are calculated for the HRV signals. Second, the high-order cumulant is estimated for the HRV and its IA, IF, and IP. Third, a high-order singular entropy is calculated to measure the fluctuation in signals. Fourth, eight novel features are obtained and processed using a DBN classifier designed for CCE prediction. The DBN classification method, with the novel HRV features, outperformed existing methods in terms of accuracy. Thus, the scheme proposed herein provided a novel direction for predicting CCEs.

Demand response is an effective way for ensuring safety and stabilization of power grid by maintaining the balance between the supply and the demand of power grid, and this article focuses on using electric water heaters for demand response. In addition to considering comfort and price factors as did in previous works, this article considers the overshoot temperature and its influence on demand response. First, a theoretical model of the heating and cooling processes of the electric water heater is established; second, the demand response process using electric water heaters is analyzed, including the influences of the physical parameters and the settings of electric water heaters on the demand response process; third, a model is established considering the demand response requirement, the comfort of owners of electric water heaters, and the electricity price, simultaneously; fourth, an optimization method based on deep reinforcement learning is proposed for demand response using electric water heaters. Meanwhile, the influence of parameters on the results of demand response is discussed in details. Experimental results show the effectiveness of the proposed method.

Obstructive sleep apnea (OSA) syndrome is a common sleep disorder and a key cause of cardiovascular and cerebrovascular diseases that seriously affect the lives and health of people. The development of Internet of Medical Things (IoMT) has enabled the remote diagnosis of OSA. The physiological signals of human sleep are sent to the cloud or medical facilities through Internet of Things, after which diagnostic models are employed for OSA detection. In order to improve the detection accuracy of OSA, in this study, a novel OSA detection system based on manually generated features and utilizing a parallel heterogeneous deep learning model in the context of IoMT is proposed, and the accuracy of the proposed diagnostic model is investigated. The OSA recognition scheme used in our model is based on short-term heart rate variability (HRV) signals extracted from ECG signals. First, the HRV signals and the linear and nonlinear features of HRV are combined into a one-dimensional (1-D) sequence. Simultaneously, a two-dimensional (2-D) HRV time-frequency spectrum image is obtained. The 1-D data sequences and 2-D images are coded in different branches of the proposed deep learning network for OSA diagnosis. To validate the performance of the proposed scheme, the Physionet Apnea-ECG public database is used. The proposed scheme outperforms the existing methods in terms of accuracy and provides a novel direction for OSA recognition.

Using unmanned aerial vehicles (UAVs) for equipment condition monitoring is an important application of Industrial Internet of Things (IIoT), and the limited energy is the key factor to restrict the application of UAV. In order to reduce the computational load for intelligence computing of UAV, this article proposes a cloud edge collaborative intelligent method for object detection, and applies it to insulator string recognition defect detection in the power IIoT. First, the impact of the extremely large aspect ratio of object on the detection accuracy and the computational load is analyzed, then the cloud edge collaborative intelligent method for insulator string detection and defect recognition is presented, in which on the UAV side a low cost method is proposed for estimating possible directions of insulator strings, and on the cloud side, an effective method is proposed for insulator string defect detection. The experimental results show the effectiveness of the proposed algorithm. To the best knowledge of us, this article is the first work to analyze the impact of the extremely large aspect ratio of insulator string on the detection accuracy and the computational load.

With the development of deep learning methods such as convolutional neural network (CNN), the accuracy of automated pulmonary nodule detection has been greatly improved. However, the high computational and storage costs of the large-scale network have been a potential concern for the future widespread clinical application. In this paper, an alternative Multi-ringed (MR)-Forest framework, against the resource-consuming neural networks (NN)-based architectures, has been proposed for false positive reduction in pulmonary nodule detection, which consists of three steps. First, a novel multi-ringed scanning method is used to extract the order ring facets (ORFs) from the surface voxels of the volumetric nodule models; Second, Mesh-LBP and mapping deformation are employed to estimate the texture and shape features. By sliding and resampling the multi-ringed ORFs, feature volumes with different lengths are generated. Finally, the outputs of multi-level are cascaded to predict the candidate class. On 1034 scans merging the dataset from the Affiliated Hospital of Liaoning University of Traditional Chinese Medicine (AH-LUTCM) and the LUNA16 Challenge dataset, our framework performs enough competitiveness than state-of-the-art in false positive reduction task (CPM score of 0.865). Experimental results demonstrate that MR-Forest is a successful solution to satisfy both resource-consuming and effectiveness for automated pulmonary nodule detection. The proposed MR-forest is a general architecture for 3D target detection, it can be easily extended in many other medical imaging analysis tasks, where the growth trend of the targeting object is approximated as a spheroidal expansion.

Continuous body temperature measurement (CBTM) is of great significance for human health state monitoring. To avoid interfering with users' daily activities, CBTM is usually achieved using wearable noninvasive thermometers. Current wearable noninvasive thermometers employ steady-state models used in nonwearable thermometers; as a result, the reaction time is long and the measurement can be disturbed by users' activities. However, there is no work to solve these issues. In this paper, first, differences between wearable and nonwearable temperature measurement are analyzed. Second, the relationship among the human body temperature, the skin temperature, and the device temperature is modeled based on artificial neural networks (ANNs). Third, this paper proposes a novel multiple ANNs-based wearable CBTM method. Experiments show that the reaction time of the proposed method is about one-tenth of that of other popular wearable noninvasive CBTM methods, while the accuracy and the robustness are improved.

Nowadays, drive recorders are becoming a popular form of evidences used by drivers and accepted by court. One common investigation task is to identify vehicles of interest and recognize their license plates (LPs). In this paper, we focus on License Plate Recognition (LPR) based on single snapshot from a drive recorder. As drive recorders are installed on moving vehicles, snapshots by drive recorders usually suffer from serious blur, and the key issue is recognizing the Blurred License Plate (BLP) from single image. A straightforward method is first deblurring the BLP and then recognizing it. However, the first problem with this method is that general image deblurring methods are designed to get a good overall visual effect and the deblurred results may be not good for LPR. The second problem is that general image deblurring methods don't use the features of the LPs, which could be important priors for the deblurring process. To overcome these issues, this paper proposes a novel method that integrates deblurring and recognizing in a closed-loop. The proposed method utilizes characters and patterns of LPs as priors, and the deblurring and recognizing process will stop when a reliable recognition result is obtained from the deblurred image. Furthermore, by analyzing the features of BLPs, this paper proposes a ℓ0-norm based deblurring method. Experiments show that, compared to other LPR methods, the proposed method can achieve higher recognition rate on the BLPs.

For a cloud based system, storage volume, users' privacy and computing capacity are three key issues. In this paper we propose a secure and effective cloud based image storage framework for e-healthcare systems with images transmission and storage. The main contribution of the proposed framework is a high compression ratio method for encrypted images. We first analyze the difficulties of encrypted image compression and discuss the drawbacks of current compressive sensing (CS) based encrypted image compression. Then we propose a novel lossy encrypted image compression method, which is based on the CS, dictionary coding, and recent sparse couple reconstruction theories. The experiment results show that compared to state-of-the-art encrypted image compression methods and classical JPEG/JPEG2000 compression methods, the proposed scheme can achieve much higher compression ratio of JPEG2000 with the similar reconstruction quality, meanwhile can obtain much better reconstruction quality than CS based methods with a similar compression ratio.

This paper addresses the issue of non-uniform motion deblurring due to hand shake for a single photograph. The main difficulty of spatially variant motion deblurring is that the deconvolution algorithm can not directly be used to estimate the blur kernel as the kernel of different pixels are different to each other. In this paper, the blurred image is considered as a weighed summation of all possible poses, and we proposed to use a PSO (particle swarm optimization) to optimize the weighed parameters of the corresponding poses after building the motion model of the camera. The main issue of using a PSO for deblurring is that it is generally impossible to obtain the ground true of the observed blurred image, which must be used as the input of the PSO algorithm. To solve this problem, firstly a novel image prediction method is proposed which combines a shock filter and a non-linear structure tensor with anisotropic diffusion. The main advantage of the proposed prediction method is that the deblurring process is not misled by rich texture in the image. Secondly an alternatively optimizing procedure is used to gradually refine the motion kernel and the latent image. Experimental results show that our approach makes it possible to model and remove non-uniform motion blur without hardware support.

Particle filtering (PF) has been widely used in solving nonlinear/non Gaussian filtering problems. Inferring to the target tracking in a wireless sensor network (WSN), distributed PF (DPF) was used due to the limitation of nodes' computing capacity. In this paper, a novel filtering method -- asynchronous DPF (ADPF) for target tracking in WSN is proposed. There are two keys in the proposed algorithm. Firstly, instead of transferring value and weight of particles, Gaussian mixture model (GMM) is used to approximate the posteriori distribution, and only GMM parameters need to be transferred which can reduce the bandwidth and power consumption. Secondly, in order to use sampling information effectively, when target moving to the next cluster head region, the GMM parameters are transfer to the next cluster head, and combine with the new local GMM parameters to compose the new GMM parameters incrementally. The ADPF can also deal with the situation of different number of nodes in different cluster when using the dynamic cluster structure. The proposed ADPF is compared to some other DPF for WSN target tracking, and the experimental results show that not only the precision is improved, but also the bandwidth and power is reduced.