In this paper, a rectangular-shaped wideband dielectric resonator Graphene inspired electronic band gap (EBG) based complementary split ring resonator (CSRR) loaded miniaturized THz MIMO antenna is designed having a dimension of 56×56×3.6μm3 is designed on a Roggers RO3035 substrate with a relative permittivity of 3.6 and loss tangent 0.0015. This antenna works in the range of 6.0THz-12.5THz (70.76%) with a peak gain of 7.68 dBi. This composite DRA materials having Silicon (εr = 11.1) based cylindrical dielectric and cylindrical Graphene disk to be stacked on rectangular radiated antenna having height of 13.5µm and 0.2µm to improve the impedance bandwidth as well as gain of the antenna from 5.24dBi to 7.68dBi. Firstly single-element gold plated rectangular shaped antenna is transformed into a two-element MIMO antenna with mushroom-shaped EBG structure is deployed for improving the isolation between two antenna elements as well as to improve the front-to-back ratio (FBR). CSRR is loaded to obtained band notch characteristics for avoiding the interference between near wireless devices in the span of 10.6-10.8THz. Graphene disk having 3µm height put on Silicon (εr = 11.1) based cylindrical DRA provides high gain and improved the impedance bandwidth for achieving wide bandwidth from 6.0THz-12.5THz. The proposed MIMO Antenna performance has been judged by antenna parameters like gain, return loss, isolation between two antenna elements, diversity parameters like ECC, DG, TARC, Mean Effective Gain (MEG), Channel Capacity Loss (CCL) and found proposed DRA-MIMO antenna suitable for wide band nano/optical communication in 6G for IoT and bio-medical sensing application.

In this paper, we propose an idea to improve various types of loss functions. It is different from the current idea of balancing the errors by increasing the number of input samples in each batch. We directly mask the top values of the error ranking to zero. During the error back-propagation, this means that the samples corresponding to that loss will not affect the parameter update of the network. In other words, even if a small number of samples are artificially mislabeled, it will not theoretically have much impact on the performance of the network. Instead, deliberately discarding anomalous losses will help smooth the training of the network. We conduct experiments on several regression and classification tasks, and the results show that the proposed method in this paper can effectively improve the expected performance of the network.

The rapidly changing software industry presents challenges in teaching software engineering concepts and skills, especially in the area of DevOps. This encompasses both technical aspects, such as pipeline automation, and non-technical ones, such as culture, principles, and team interaction modes. This presents a significant challenge and there is high demand for skilled DevOps practitioners. To address these challenges, we propose a teaching strategy for DevOps that is based on real-world experiences from companies adopting DevOps and real projects. The strategy incorporates two teaching techniques: Research-based Teaching and Open Source Software projects. Our findings can help educators effectively plan, implement, and evaluate DevOps courses. The results also highlight several opportunities for practitioners to benefit from academic environments and experiment with new ideas to train abilities expected in new hires.

The diagnostic approach of interstitial lung diseases (ILDs) may be a challenge and requires a multidisciplinary discussion (MDD). While the definition of histological patterns of surgical lung biopsy by pathologists is helpful to the final diagnosis of MDD, its recognition could be challenging due to various degrees of lung remodeling and distortion. The pleura is a reference structure to identify some histological findings for the recognition of a pathological ILD pattern. When a pattern is found, it is important to know whether it is close to the pleura to determine its specific type and severity. This manual process is a tedious and laborious one for the pathologist. Automating this task is important for a complete computed-system-assisted ILD diagnostic process. We introduce “RadPleura” a framework for pleura classification of histopathological images using a radiomics-based approach. Our framework performs image pre-processing, region-of-interest segmentation, and extraction of a radiomic-signature fitted to  ILD classification. To evaluate the radiomic-signature, we classified it into pleura and non-pleura, using two classifiers: Support Vector Machine and Gradient-boosted Decision Trees. Our experiments yield promising results, with F-scores of 92% for SVM, and 91% for GBD. We also created a dataset of lung histopathology images with respective ground truth for pleura classification. To the best of our knowledge, this study is the first disclosed attempt to explore and develop a radiomic-signature for pleura classification. The methods and techniques are integrated into the RadPleura framework developed.

In intelligent reflecting surface (IRS)-assisted communications, the ultimate gain is achieved when the phases of the reflected signals are optimally selected to maximize the signal-to-noise ratio (SNR). However, practical hurdles, particularly the imperfect phase estimation and quantization can reduce the potential gain. Therefore, this work aims at evaluating the impact of applying a quantized phase in the presence of phase estimation errors. Towards this goal, we derive the probability density function (PDF) of the estimated quantized phase, then using the sinusoidal addition theorem (SAT), the PDF of the received signal envelope is derived and used to derive closed-form expressions of the symbol error rate (SER) and outage probability (OP). The obtained analytical and simulation results show that the SER and OP jointly depend on the SNR, phase estimation accuracy, number of IRS elements, and number of quantization levels. The imperfect phase and quantization demonstrated several counterintuitive results. In particular, it is shown that increasing the number of IRS elements or the number of quantization levels may degrade the system performance. Moreover, the results reveal that the impact of phase quantization increases as the phase estimation accuracy decreases. The results also show that the performance is susceptible to phase errors with an even number of reflectors and binary quantization levels.

Radiation detection systems for security applications are often equipped with neutron sensors and required to promptly respond with an alarm if the neutron flux exceeds the background level. As the ambient neutron flux is essentially caused by cosmic radiation, this background could be evaluated by assessing the rate of muons crossing the gamma-ray detector which is comprised in such systems anyway. A study performed with a commercial Backpack Radiation Detection System (BRD) demonstrates the feasibility of this approach, supposed the gamma spectrometer is capable of discriminating muon signals from energetic events caused by the gamma-ray cascades following neutron captures. This requires a dynamic range exceeding the usual 3-4 MeV. The instrument used provided standard spectroscopy up to 10 MeV and facilitated charge-loss compensated spectroscopy up to 1 GeV, which allowed a separate assessment of normal gamma events (< 3 MeV), neutron-capture gamma events (3-7 MeV), and muon events (> 10 MeV). In this case, the rate of neutron-capture gamma events can be used as a neutron indicator on its own, while the muon rate signals the ambient neutron background level.

A document by Qiang Zhu . Click on the document to view its contents.

Requirement engineering is mostly concerned with communicating with customers to gather the essential and relevant domain information that forms the base of requirements and which is the key area for research on how to gather and record information from stakeholders (customers, end-users, domain experts). The main obstacle here is the ambiguity of Natural Language (NL). In this research, we present a user-centered process by applying a process model for creating and deriving natural language requirements in several steps/workshops. The process model was supported by examples of preventing natural language defects, applying templates, and helping to use logical decomposition to derive technical requirements.

Intelligent reflecting surface (IRS) is considered as an enabling technology for millimeter wave (mmWave) communications since it can provide an extra reflection link to increase the macrodiversity gains and improve the transmission performance. These advantages principally rely on the assumption that the blocking of direct and reflection links is mutually independent. However, due to the nonnegligible sizes of blockages in some practical cases, both links may be simultaneously blocked by the same blockage, which is the so-called blockage correlation. To this end, in this work we investigate the impact of blockage correlation in the IRS-assisted mmWave communication systems. Firstly, we provide a new joint line-of-sight/non-line-of-sight (LOS/NLOS) probability model with blockage correlation. The correlation coefficient of direct and reflection link states is also derived. Secondly, we study the effects of blockage correlation on the system performance of IRS-assisted mmWave communication systems by deriving the expression of transmission success probability. Moreover, we provide a deployment optimization algorithm for IRS when blockage correlation is considered. Finally, simulations verify our theoretical results and validate the IRS deployment optimization algorithm. It is shown that the widelyused independent blocking assumption not only always incur an overestimation of the performance of IRS-assisted mmWave communication systems, but also misdirect the optimal deployment of IRS.

We present a model-agnostic federated learning method for decentralized data with an intrinsic network structure.The network structure reflects similarities between the (statistics of) local datasets and, in turn, their associated local models. Our method is an instance of empirical risk minimization, using a regularization term that is constructed from the network structure of data. In particular, we require well-connected local models, forming clusters, to yield similar predictions on a common test set. In principle our method can be applied to any collection of local models. The only restriction put on these local models is that they allow for efficient implementation of regularized empirical risk minimization (training). Such implementations might be available in the form of high-level programming frameworks such as scikit-learn, Keras or PyTorch.

Study of random propagation of malware in Internet-of-Things (IoT) networks. The study focuses on deriving exact mathematical models for the random propagation of malware using Markov chains and concepts of epidemic compartmental models.

A numerically efficient method of shooting and bouncing rays for sparse particles (SBR-SP) is presented to analyze a large two-dimensional (2D) array of magnetodielectric circular cylinders which are distributed with any positions, sizes, and media on the xy-plane. Iterative- and closed-forms of the SBR-SP are used to understand the convergence and accuracy characteristics of a large array of cylinders and we also compare them with the results obtained by a mode-matching technique, thus resulting in favorable agreement. Especially, the closed-form of the SBR-SP with GPU acceleration is fast and accurate enough to obtain various echowidth behaviors of a 80×80 square array  within  71.8 s. In addition, a statistical analysis for a randomly distributed array with 6,400 elements is conducted in terms of mean and standard deviation and we obtain that initial shooting rays excited by an incident field are strongly dominant in statistical behaviors of scattering patterns. Our algorithm based on the SBR can be extended to three-dimensional (3D) problems where we model a large array of lossy dielectric spheres for the estimation of rain attenuation and signal variation.

The origin of electromagnetism has been a recurring question since the discovery of electricity. In the 19th century, Maxwell unified electricity and magnetism into a single electrodynamic theory described by Maxwell’s equations. This was followed by the discovery of the electrodynamic force on a moving charge by Lorentz. Later, Einstein demonstrated that magnetism and the Lorentz force can be considered a relativistic effect. However, despite these important theories, there is at present no physical theory that identifies the single fundamental cause of all known classical electromagnetism. In an attempt to identify this single fundamental cause, this paper shows that classical electromagnetism can be considered as effect of scalar waves emanating from charges. It is shown in this paper that Coulomb’s law, Maxwell’s equations, the magnetic and electric field, the Lorentz force, and the Lienard-Wiechert potentials, can all be derived only from this single hypothesis that scalar waves are emitted by the charges. In addition, it was shown that the Lorentz transformation can also be derived from this hypothesis of scalar waves. To achieve this, an important new quantity was introduced: the momentum of the scalar waves. It was shown that the force caused by these scalar waves is simply the time derivative of this momentum. It was also shown that both the Lorentz force and the Coulomb force can be represented as the time derivative of the scalar wave momentum.

Our study aims to detect multilingual Out-of-Vocabulary (OOV) and matching among multilingual OOV. Based on the original OOV issue, many multilingual OOVs also emerged at the same time. In order to solve this problem, this paper proposes a graph embedding-based matching among multilingual OOV. The method is divided into two parts. The first part is to extract OOV from the network corpus and understand it. In the second part, the OOV in the first part is taken as the target node, and the understood part is taken as the feature node of the target node to construct the graph and embed the graph. Our study uses Chinese, Korean, and Japanese for the experiment. The result of the method that Our study proposed is that F1- score reached 93.94%. Our study also compares this method with other embedding algorithms, and the F1-score is higher than the average of other algorithms F1-score by 9.62%.

An Artificial Neural Network (ANN) based Optimal Feedforward Torque Control (OFTC) strategy for electrically excited synchronous machines (EESMs) is proposed. After design, data set creation, training and validation of the ANN, the analytical computation of the optimal stator and exciter currents is achieved which allows to minimize copper and iron losses and to produce the desired (or maximally feasible) machine torque. Voltage and currents constraints of stator and exciter are considered as well. In contrast to conventional OFTC, the proposed ANN-based OFTC strategy does not require iterations nor a decision tree to find the optimal current triple while machine nonlinearities, magnetic cross-coupling, saturation and speed-dependent iron losses are taken into account. In addition, the proposed ANN design procedure allows to consider measurable OFTC goals and computational resources that ensures a real-time capable implementation. Comprehensive simulation results for a real and nonlinear EESM clearly show these benefits by comparing the proposed ANN-based OFTC with results of a nonlinear optimization problem (NLP) solver (which cannot be used in real-time).

This paper proposes a novel data-driven approach to designing orthonormal transform matrix codebooks for adaptive transform coding of any non-stationary vector processes which can be considered locally stationary. Our algorithm, which belongs to the class of block-coordinate descent algorithms, relies on simple probability models such as Gaussian or Laplacian for transform coefficients to directly minimize with respect to the orthonormal transform matrix the mean square error (MSE) of scalar quantization and entropy coding of transform coefficients. A difficulty commonly encountered in such minimization problems is imposing the orthonormality constraint on the matrix solution. We get around this difficulty by mapping the constrained problem in Euclidean space to an unconstrained problem on the Stiefel manifold and leveraging known algorithms for unconstrained optimization on manifolds. While the basic design algorithm directly applies to non-separable transforms, an extension to separable transforms is also proposed. We present experimental results showing that adaptive coding with our transform codebook designs outperform the commonly used discrete cosine transform (DCT) in many regions of natural images and motion-compensated prediction error images of video frames. According to these results, the proposed codebook designs slightly outperform those found by a common alternative approach, the sparsity-based optimization.

Software quality definition encompasses six key factors. These are low levels of defects, high reliability, efficient structure that minimizes insertion of new defects during defect repairs, effective customer support, high user satisfaction, and a high defect repair rate, especially for high-severity defects. Several methods have been proposed for measuring software quality. These include fuzzy logic based frameworks, neuro fuzzy models, unsupervised learning methods, queuing models, and software quality metrics. This research paper reviews different frameworks, models, metrics, and evaluation techniques that have been published in various research papers for estimating and assessing software quality. Research publications related to software quality measurement, estimation, and assessment were collected for the literature review. These publications were accessed from online research databases that consist of open access (OA) journals.

Synchronous machines (SMs) like reluctance synchronous machines (RSMs) or interior permanent magnet synchronous machines (IPMSMs) are characterized by nonlinear stator flux linkages which are crucial for optimal control and operation management. This paper presents a novel disturbance observer that allows to estimate the nonlinear flux linkages online for any SM (with constant or no excitation). Moreover, a simple identification sequence with ramp-like voltage signals is proposed to extract key information of the nonlinear stator flux linkage maps within a short period of time and without the need of any controller, torque sensor or prime mover. Finally, the proposed observer with identification sequence is implemented and validated for a real and highly nonlinear IPMSM by realistic and detailed simulation results.