Vol. 45 · No. 4

IIoT devices are the foundation of the Industrial Internet of Things (IIoT), including functions such as environmental perception, data processing, and remote operation. One core advantage of IIoT devices is their remote operation capability, allowing operators to control them via smart devices from distant locations. While remote operation greatly enhances flexibility, the resulting security issues must not be overlooked. Instructions can be tampered with during transmission. Existing devices can verify the sender's identity, but a legitimate sender may still issue non-compliant instructions, impacting safety and efficiency of industrial production. Most existing access control schemes for IIoT devices are unsuitable for secure remote operations. They focus on data exchange between devices, ignoring compliance of data content and sender. This paper proposes a policy-hiding attribute-based access control scheme for secure remote operation of IIoT devices. It provides fine-grained access control while ensuring safe operation by verifying sender compliance and instruction content through a compliance policy. Additionally, policy hiding protects sender privacy. Furthermore, multi-receiver signcryption is employed to prevent unauthorized tampering with the instructions during transmission. The security analysis substantiates the security of our scheme, while the performance analysis demonstrates its applicability in IIoT scenarios.

This study aims to compare the behavioral outputs and response durations' analysis of two distinct groups of medical professionals – general practitioners and orthopedic specialist doctors – utilizing deep learning and computer vision algorithms. This comparison is facilitated through one-way ANOVA and Tukey pairwise comparison tests. The research focuses on femoral proximal fractures, which have the highest disability and mortality rates among the elderly population globally. For this research, over 1 500 femoral proximal region images sourced from over 500 patients aged between 22 and 105 for retraining of deep learning algorithms. Updated SSD – MobileNet v2, Updated Faster R-CNN – Inception v2 and Updated YOLO Darknet v4 algorithms were retrained specifically for femoral proximal fracture detection. In this study, a total of 50 X-ray images comprising femoral proximal fractures and a total of 50 X-ray images without fractures were gathered, in total evaluation dataset containing 55 X-ray images, underwent scrutiny via one-way ANOVA tests, assessing both tree retrained computer vision deep learning algorithms, as well as physician groups concerning fracture detection success in fractured and non-fractured regions, for the entirety of the evaluation dataset. Subsequent Tukey pairwise comparison tests were conducted based on the obtained results. Regarding result production durations, it was observed that the average detection time of expert physicians significantly exceeded that of general practitioners, the Updated YOLO algorithm, the Updated Faster R-CNN algorithm, and Updated SSD Algorithm. Conversely, no significant disparity was noted between the average result production times of general practitioners and expert physicians. In Tukey pairwise tests evaluating the accuracy rate of fractured regions, the success rate of expert physicians in fractured region accuracy significantly surpassed that of SSD and general practitioners, while insignificantly differing from Faster R-CNN and YOLO Algorithms. For non-fractured region accuracy, the Tukey pairwise test revealed that the accuracy rate of the YOLO algorithm significantly outstripped that of Faster R-CNN and SSD Algorithms, while insignificantly differing from expert and general practitioners. However, it was noted that the success rate of Faster R-CNN and SSD algorithms significantly differed from expert physicians. In conclusion, individual output behavior differs between general practitioners and expert physicians, but expert physicians show (like AI Algorithms) similar behaviors in the same outputs.

Digital data and automated solutions have raised the standard of medical image analysis and made it more complex and challenging. Manual examination of medical images is inefficient, subjective, and error-prone due to variation in attributes such as shape, size, and texture. The majority of approaches in the literature, starting from the classical U-Net including its modern variants, usually suffer from overfitting issues, model saturation issues, and inability to capture multi-scale contextual features, reducing generalization across different imaging modalities. To address these challenges, we present ResCAM-Net, an improved encoder–decoder architecture that accurately and efficiently raises segmentation accuracy for different types of medical images. Our model incorporates residual learning, multi-kernel residual convolutions, adaptive feature recalibration using atrous spatial pyramid pooling, and a hybrid triple attention module to enhance feature aggregation and focus on critical regions. ResCAM-Net reduces the number of trainable parameters to 9.91 million, compared with other state-of-the-art architectures, by approximately 70

Multi-modal 3D object detection, which integrates point clouds and images, has gained increasing attention due to its potential to enhance detection accuracy. However, challenges such as redundant information and scale inconsistencies hinder effective feature fusion. To address these issues, we propose DMS3D (Multi-modal 3D object detection algorithm based on discriminant fusion strategy and multi-scale feature fusion), a Discriminative fusion strategy and Multi-scale feature fusion-based System for 3D object detection. Our method introduces two key modules: 1) a multi-scale feature fusion module (MSFF) based on graph neural networks, which refines point cloud representations by aligning them with image semantics while preserving fine-grained details; and 2) a discriminative fusion module (DFM) with a lightweight gating mechanism, which selectively fuses multi-modal features to reduce redundancy and noise while enhancing computational efficiency. Extensive experiments on the KITTI and nuScenes datasets validate the effectiveness of DMS3D, demonstrating superior detection performance compared to state-of-the-art methods.

The two biggest problems facing modern blockchain systems are scalability and privacy. The system's capacity to scale is constrained by the need to process transactions at each node. In addition, despite all other benefits, a widely cited barrier to blockchain adoption is the requirement to disclose all transaction data at every node, thereby making the data public. In order to overcome these two issues, we introduce the ScaZK (Scalable Zero Knowledge) protocol, which primarily makes two contributions: i) In order to improve performance, we propose a new, lightweight protocol that uses non-interactive proofs for off-chain calculations, which limits on-chain computational efforts to the verification of execution accuracy rather than actual execution. Private data utilised in the off-chain computation does not need to be made public in order to verify correctness because of the verifiable computation scheme's zero-knowledge property. ii) We present a framework for defining, integrating, and deploying these off-chain calculations. It includes a compiler, proof and verification generators, and a domain-specific language for smart contracts. It hides the considerable complexity that zero-knowledge proofs entail, offers developers a more comfortable and advanced level of programming abstractions, and allows circuit integration, which promotes adoption. Our contributions in this work specifically include: i) a novel proof bucketing compression scheme, ii) integration of a B+ Merkle Tree for efficient rollup root computation, iii) parallel off-chain execution for throughput scalability, and iv) partial zkEVM compatibility for developer accessibility. We also present stress testing, fault recovery analysis, and gas breakdown for multiple proof systems to validate the protocol's resilience and scalability.

This paper presents the development and implementation of a teleoperation system using motion capture methodologies for the Baxter robot. The system utilizes a Microsoft Kinect V2 camera to capture human motion data and translates it into robot joint angles in real time, enabling intuitive control of the robot's movements. A function approximation technique (FAT) based controller was implemented to enhance the system's stability and performance while compensating for unknown dynamics. The mathematical framework incorporates the Denavit-Hartenberg model for mapping human arm movements to the robot's 7-DOF structure, along with space vector calculations for joint angle determination. Experimental results demonstrate successful real-time motion replication across multiple joint configurations, with the robot effectively mimicking human arm movements. Data analysis of joint positions, velocities, and accelerations confirms the system's capability to maintain accuracy while managing inherent challenges such as network latency and motion jittering. The developed system proves efficient and cost-effective, utilizing readily available hardware components while achieving reliable performance. This work contributes to advancing human-robot interaction in applications where direct human presence may be hazardous or impractical, offering potential benefits across industries including manufacturing, healthcare, and hazardous environment operations.

Continuous glucose monitoring is essential for effective diabetes management, but most clinically validated systems rely on minimally invasive sensors that require subcutaneous insertion, periodic calibration, and frequent replacement. These limitations create a critical need for non-invasive glucose estimation methods that are accurate, wearable, and deployable on resource-constrained devices. In this study, we propose a lightweight neural network architecture for non-invasive glucose estimation using biomedical signals, including PPG, ECG, and accelerometer (ACC) signals. To address the non-stationary nature of physiological signals, we employ discrete cosine transform (DCT) representations that preserve both temporal and spectral characteristics. The model uses a multi-branch convolutional architecture with anisotropic kernels that independently capture temporal dynamics and spectral correlations, enabling efficient feature extraction while having only 20.5 k parameters. Evaluation on two public datasets demonstrates robust cross-modal performance, achieving over 98

RoCE is widely deployed in data centers for AI training, but network congestion causes GPU underutilization and slows training. Existing congestion control algorithms suffer from slow feedback loops and isolated handling of slowdown and recovery, leading to oscillations and poor bandwidth utilization. We propose CHOCOLATE, which integrates reverse explicit notification (REN) with lightweight flow prediction for congestion sensing, and unifies slowdown and recovery into a coordinated control system. Experimental results demonstrate improved RoCE performance in lossy networks.

The designation of milestone activities in business processes can restrict the effective occurrence of cases in logs and remove noise reasonably. A deviation detection method is presented between Petri nets and logs using milestone activities to improve computational efficiency. An original model is constructed through some operations between two Petri nets, where one is the process model and the other is the log model created by mapping the activities in the case to the transitions. First, the transitions labeled with the milestones compose the synchronous transitions by synchronous composition. Next, other transitions with the same labels produce the log, model and synchronous transitions by product. Then, the transitions labeled with the activities that are merely observed in the model directly map to the transitions. Finally, the transitions labeled with the activities which are modeled only in the Petri net directly map to the model transitions, and the transitions labeled with no activities directly map to the invisible transitions. The new model is the search space for deviation detection. An algorithm is presented to detect the deviations between the Petri net based on the given cost and the case. The experimental results show that the more milestone activities in the business process, the higher computational efficiency of deviation detection.