Search

With the development of Consumer Electronic Products (CEP) in the Computer, Communication, and Consumer electronics (3C) industry, research over high speed, high precision, and reliability CEP has been an urgent need. Due to the high manufacturing requirements, it has been challenging to product, transmit, assembly, or test the CEP. Metaverse with Virtual Reality (VR)/Augmented Reality (AR) technologies connect reality and virtual, which provides a more convenient manufacturing platform for CEP, thus bringing up the concept of Metaverse Consumer Electronic Products (MCEP). Digital Twins (DT) has effectively provided an emulated software replica to create the Metaverse. DT for MCEP (DTM) facilitates the implementation of more advanced applications, especially Artificial Intelligence (AI) applications. An essential component in DTM AI applications is the labeled dataset, but such carefully labeled data are usually inaccessible since labeling is costly. In addition, the unlabeled data are usually stored locally, referred to as Federated Learning (FL). Previous works adopted Domain Adaptation to construct a pseudo-labeled dataset. However, those approaches have rising potential in accuracy, convergence speed, and communication efficiency. In this paper, we propose cFedDT, a cross-domain FL algorithm, to construct a pseudo-labeled dataset in DTM. cFedDT contains three modules. Ensembled Knowledge for Cross Distillation distills the knowledge from the source models to the target model to improve the target accuracy. Adaptive Batch adjustment for Stable Convergence gradually adjusts the number of batches on each client to speed up the converging process. Magnitude Pruning for Efficient Communication reduces the parameters that require communication to speed up the training. Our experimental results illustrate the priority of cFedDT. Compared with other state-of-the-art algorithms, cFedDT can increase the target accuracy by 9.8% or reduce up to 99.97% of the communication with a target accuracy loss of less than 2.7%.

With the increase in data volume and environment complexity, real-world problems require more advanced algorithms to acquire useful information for further analysis or decision making. Cognitive learning (CL) effectively handles incomplete information, and multiagent systems can provide enough data for analysis. Inspired by distributed machine learning, federated learning (FL) has become an efficient framework for implementing CL algorithms in multiagent systems while preserving user privacy. However, traditional communication optimizations on the FL framework suffer from either large communication volumes or large accuracy loss. In this article, we propose pFedEff, a personalized FL framework with efficient communication that can reduce communication volume and preserve training accuracy. pFedEff uses two magnitude masks, two importance masks, and a personalized aggregation method to reduce the model and update size while maintaining the training accuracy. Specifically, we use a pretraining magnitude mask for approximated regularization to reduce the magnitude of ineffective parameters during training. We also use a post-training magnitude mask to eliminate the low-magnitude parameters after training. Furthermore, we use uploading and downloading importance masks to reduce the communication volume in both upload and download streams. Our experimental results show that pFedEff reduces up to 94% communication volume with only a 1% accuracy loss over other state-of-the-art FL algorithms. In addition, we conduct multiple ablation studies to evaluate the influence of hyperparameters in pFedEff, which shows the flexibility of pFedEff and its applicability in different scenarios.

With the rapid development of wireless communication, achieving the neXt generation Ultra-Reliable and Low-Latency Communications (xURLLC) in 6G mobile communication systems has become a critical problem. Among many applications in xURLLC, deep learning model inference requires improvement over its efficiency. Due to the heterogeneous hardware environment in 6G, parallel schedules from distributed machine learning and edge computing has been borrowed to tackle the efficiency problem. However, traditional parallel schedules suffer from high latency, low throughput, and low device utility. In this paper, we propose Automatic Pipeline Parallelism ( $AP^{2}$ ), a parallel inference framework for deep learning applications in 6G mobile communication systems, to improve the model inference efficiency while maintaining reliability. $AP^{2}$ contains three sub-modules. A task-device affinity predictor predicts a task’s expected execution time on a given device. The parallel inference arrangement optimizer finds the most suitable device for each task. The parallel inference scheduler converts the arrangement to a schedule that can be directly executed in the system. The experimental results show that $AP^{2}$ can achieve better latency, throughput, reliability, and device utility than other parallel schedules. Also, the priority of the sub-module designs has been approved through the experiments.

The ultralow frequency oscillation damping problem caused by asynchronous networking mode is investigated for Chongqing-Hubei asynchronous grid. The negative damping problems offered by governors are investigated. The electromechanical transient model of the voltage source converter (VSC)-high voltage direct current (HVDC) for the Chongqing-Hubei asynchronous grid is constructed. Fault modes that cause ultralow frequency oscillation after VSC-HVDC are analysed by using the Power System Analysis Software Package (PSASP), because faults in asynchronous grid often result in the ultra-low frequency oscillation damping. An additional damping controller of the VSC-HVDC and its small signal model is proposed, and the mechanism of the additional controller which increases the system damping is also analysed. The simulation results show that low frequency oscillation can be effectively suppressed by using the provided controller in this paper.

The Fifth Generation (5G) mobile systems support many kinds of intelligent cloud services. The upcoming Sixth Generation (6G) mobile systems aim to provide customized cloud services since the users have different capabilities and needs. However, as an essential scenario in 6G mobile systems, collaborative AI failed to deliver customized cloud services to the users because of system heterogeneity. Due to the lack of efficient resource coordination and service orchestration strategy, collaborative learning suffers from high latency and low resource utilization. This article proposes Hierarchical Adaptive Collaborative Learning, a distributed learning framework for customized cloud services in 6G mobile systems, to improve the training efficiency and resource utilization by dynamically adjusting the collaborative training process according to the service and resources. We introduce the critical scenarios and challenges of distributed learning. Then we present our framework to handle heterogeneous cloud services and resources. Next, we illustrate various model training and collaborating techniques to improve training efficiency. Finally, we provide a case study to show the priority of the modules in our framework and provide future research directions on collaborative learning for customized cloud services in 6G mobile systems.

A mathematical description of the absolute surface height distribution in generalized fringe projection profilometry under large measuring depth range is presented. Based on least-squares polynomial fitting, a nonlinear calibration to determine the mapping between phase change and surface height is proposed by considering the unequal height arrangement of the projector and the camera. To solve surface height from phase change, an iteration method is brought forward. Experiments are implemented to demonstrate the validity of the proposed calibration and an accuracy of 0.3 mm for surface profile under 300 mm measuring depth can be achieved.

In fringe projection, the CCD camera and the projector are often placed at equal height. In this paper, we will study the calibration of an unequal arrangement of the CCD camera and the projector. The principle of fringe projection with two-dimensional digital image correlation to acquire the profile of object surface is described in detail. By formula derivation and experiment, the linear relationship between the out-of-plane calibration coefficient and the y coordinate is clearly found. To acquire the three-dimensional (3D) information of an object correctly, this paper presents an effective calibration method with linear least-squares fitting, which is very simple in principle and calibration. Experiments are implemented to validate the availability and reliability of the calibration method.

Measuring deformation of vibrating specimens whose dimensions are in the submillimeter range introduces a number of difficulties using laser interferometry. Normal interferometry is not suitable because of a phase ambiguity problem. In addition, the noise effect is much more serious in the measurement of small objects because a high-magnification lens is used. We present a method for full-field measurement of displacement, velocity, and acceleration of a vibrating miniature object based on image-plane digital holographic microscopy. A miniature cantilever beam is excited by a piezoelectric transducer stage with a sinusoidal configuration. A sequence of digital holograms is captured using a high-speed digital holographic microscope. Windowed Fourier analysis is applied in the spatial and spatiotemporal domains to extract the displacement, velocity and acceleration. The result shows that a combination of image-plane digital holographic microscopy and windowed Fourier analyses can be used to study vibration without encountering a phase ambiguity problem, and one can obtain instantaneous kinematic parameters on each point.

Abstract: Objective To segment and measure the upper airway using cone-beam computed tomography (CBCT). This information may be useful as an imaging biomarker in the diagnostic assessment of patients with obstructive sleep apnea and in the planning of any necessary therapy. Methods With Institutional Review Board Approval, anonymous CBCT datasets from subjects who had been imaged for a variety of conditions unrelated to the airway were evaluated. DICOM images were available. A segmentation algorithm was developed to separate the bounded upper airway and measurements were performed manually to determine the smallest cross-sectional area and the anteriorposterior distance of the retropalatal space (RP-SCA and RP-AP, respectively) and retroglossal space (RG-SCA and RG-AP, respectively). A segmentation algorithm was developed to separate the bounded upper airway and it was applied to determine RP-AP, RG-AP, the smallest transaxial-sectional area (TSCA) and largest sagittal view airway area (LCSA). A second algorithm was created to evaluate the airway volume within this bounded upper airway. Results Measurements of the airway segmented automatically by the developed algorithm agreed with those obtained using manual segmentation. The corresponding volumes showed only very small differences considered clinically insignificant. Conclusion Automatic segmentation of the airway imaged using CBCT is feasible and this method can be used to evaluate airway cross-section and volume comparable to measurements extracted using manual segmentation.

PurposeSegmentation of the maxillary sinuses for three-dimensional (3D) reconstruction, visualization and volumetry is sought using an automated algorithm applied to cone beam computed tomographic (CBCT) data sets.PurposeSegmentation of the maxillary sinuses for three-dimensional (3D) reconstruction, visualization and volumetry is sought using an automated algorithm applied to cone beam computed tomographic (CBCT) data sets.