In this paper, the design, manufacture and performance test of inductors based on FeCoB thin films are studied. Firstly, FeCoB films were prepared by physical vapor deposition. Then, inductors with different turns were prepared on the substrate by micromachining technology. Finally, the inductor characteristics are characterized by microwave detector and vector network analyzer. The results show that the inductance and quality factor of the prepared thin film inductors are greatly improved and the self-resonant frequency is maintained.
A parasitic metal post scheme is proposed to enhance the cross-polarization discrimination (XPD) of $\pm 45^{\mathrm{o}}$ dual-polarized cross-dipole antennas in this paper. The technique's effectiveness is verified by designing vertical and horizontal $\pm 45^{\mathrm{o}}$ dual-polarized antenna elements. The array-antenna decoupling surface (ADS) decoupling scheme and the proposed XPD enhancement scheme are applied to four $2\times 2$ arrays to explore the relationship between enhancing XPD and decoupling. The results show that decoupling and enhanced XPD are independent, due to which the design of large-scale base station (BS) arrays can be convenient in future methodologies.
This paper reports $a 180^{\mathrm{o}}$ hybrid coupler with filter function. Comprising four resonators, the proposed coupler achieves a $\boldsymbol{180^{\mathrm{o}}}$ phase shift through different type of couplings. The whole structure is analyzed theoretically, and it is proved that the structure can realize the filter $180^{\mathrm{o}}$ hybrid coupler. To test the correctness of this design method, a $\boldsymbol{180^{\mathrm{o}}}$ hybrid coupler with second-order filtering characteristics is devised, manufactured and tested. The coherence observed between the simulation and experimental results substantiates the accuracy and reliability of the design methodology.
In order to improve the positioning accuracy and reliability of intelligent farm machinery in orchard environment, a combined GNSS(Global Navigation Satellite System)/SINS (Inertial Navigation System)navigation system for orchard in loose combination mode was built based. An improved UKF algorithm, IUKF (Improved Unscented Kalman Filter), is proposed for the problem that interference in the orchard environment causes observation coarseness in the measurement data and the standard UKF causes degradation in the filtering accuracy of the combined navigation system. Physical experiments show that compared with UKF, the average absolute error of IUKF is reduced by more than 47.16% and the root mean square error is reduced by more than 59.78%, and IUKF meets the requirements of orchard positioning accuracy.
Due to the increasing risk of data security, distributed learning model based on real-world data analytics has attracted more attention, and it has been applied in a variety of areas ranging from medical screening to agriculture, industry, finance, and defense science. Generally, participants provide their own private datasets to efficiently train the distributed models on real-world data, which inevitably leads to privacy and security concerns. Without uploading raw training data, Federated Learning enables large-amount nodes to train a distributed model and preserves security and privacy of user sensitive information. However, Federated Learning is limited by expensive computational costs during collective parameter server aggregation. Moreover, malicious nodes among computing nodes interfere model training to some extent and further cause the leakage of data privacy. To address the above-mentioned problems, we propose a novel Decentralized Federated Learning by integrating blockchain and federation learning for efficient node selection and communication. Based on the proposed model, a reputation-based learning nodes selection algorithm is presented to measure the probability of honest participation of distributed nodes. The simulation results demonstrate that our RBLNS is capable of improving the training result significantly and decreasing convergence time.
The paper introduces a large cavity Vivaldi antenna with diverters for breast tumor detection. The antenna size is 45×40×1.2 mm 3 , A 2.5-10 GHz bandwidth was achieved by adjusting parameters such as index lines and cavity radius. Simulation results confirm that the antenna satisfies ultra-wideband features. Subsequently, we designed a microwave imaging experiment, and the antenna correctly located the tumor, demonstrating the antenna's excellent performance.
In this work, we propose a no-reference image quality evaluation approach, aiming to solve the problem that the traditional convolutional neural network is insufficient to express the global information of the image. Therefore, in order to make the captured image features have the coherence of global context information, we combine a Transformer structure widely used in NLP (natural language processing) with the traditional CNN to realize the attention mechanism in the human visual perception characteristics. Firstly, the self-attention mechanism of Transformer is used to learn the global representation of the image from the multi-layer features extracted from different levels of CNN. Additionally, to forecast the ultimate image quality score, the global and local attributes are combined. Finally, the proposed algorithm is evaluated on three public datasets commonly used in CSIQ [1], TID2013 [2] and LIVE [3]. Compared with eight traditional no-reference image quality evaluation algorithms and five methods based on deep learning, the proposed algorithm achieves competitive results.
In order to improve the handset receiver performance, a wideband circularly polarized antenna is designed in this paper. A truncated square patch is used as a driving element to generate circularly polarized wave. Then a non-uniform metasurface structure is used to obtain the wider band; and the characteristic mode analysis is performed on the structure. By loading trapezoidal base and ‘Feng’ font matching branch, the surface wave on the non-uniform metasurface can be better excited. Thus, the axial ratio bandwidth and the impedance bandwidth have been improved efficiently. Experiment results show that the designed antenna is with a good left-handed circularly polarized radiation property; and it can be used in satellite communication.
A broadband low-noise amplifier (LNA) designed for the 8-10GHz wireless communication frequency range was developed based on the 180nm CMOS process. The overall circuit adopts a two-stage self-biasing common source-common gate (cascode) structure to improve gain. A parallel feedback circuit is employed to extend the bandwidth, and gain compensation techniques are utilized for interstage matching to improve gain flatness. The simulation results show that within the range of 8-10GHz, the gain is greater than 26.63dB, a flatness fluctuation of $\boldsymbol{\pm 0.44}\mathbf{dB}$ , and a maximum noise figure below 2.78dB. The power consumption is 18.2mW, and the chip area is $\boldsymbol{529.21} \mathbf{um} \times \mathbf{5 8 9. 5 3 u m}$ include RF pad.
In practical applications, noise is more appropriately modelled as an a-stable random process than as a Gaussian process. To cope with a-stable noise, a fractional-order stochastic gradient descent (SGD) method is invented to adaptive filtering (AF) algorithms. When both input signal and noise follow α-stable random processes, the AF algorithms based on mean square error (MSE) criteria can be affected. Nevertheless, the fractional-order AF (FoAF) algorithm can effectively address this issue. For the purpose of better stability performance of the FoAF algorithm, different cost functions (CFs) are introduced into fractional-order SGD method. Based on the fractional-order derivative of a logarithmic CF (LCF), a fractional-order logarithmic adaptive filtering (FoLAF) algorithm is formed in this condition. Trials show that the FoLAF algorithm achieves faster convergence, excellent steady-state performance and superior tracking performance.
The normalized subband adaptive filtering algorithm provides fast convergence rate for colored input signals as compared to the normalized least mean square algorithm, but it suffers from a poor convergence issue in the α -stable noise. In light of this, the normalized subband p-norm (NSPN) algorithm, which is based on the least mean p -power error (MPE) criterion, is proposed in this study. This technique is not only robust against impulsive noise samples, but it also maintains a fast convergence rate when colored input signals are used. In addition to this, we develop both the steady-state and the transient models of the NSPN algorithm and provide some insights. Then, in order to solve the problem of making a choice regarding the order p in the NSPN algorithm, we design an autonomous system and come up with the NSPN algorithm with a variable p-norm (VP-NSPN). In addition, we offer the TFC-based VP-NSPN algorithm with a fast convergence rate and low steady-state misadjustment simultaneously by making use of the tap-weights feedback-based convex combination (TFC) scheme. In conclusion, simulation results on system identification and acoustic echo cancellation are undertaken in order to validate the superiority of the proposed algorithms and validate the usefulness of the theoretical analysis.
Low-light images will likely suffer from multiple issues, such as color distortion, excessive enhancement, halo artifacts, and noise amplification when their brightness is increased. The existing low-light image enhancement methods are challenging to solve the above problems simultaneously, especially with poor performance in dealing with halo artifacts and noise. In this paper, we propose a comprehensive method that can simultaneously solve multiple challenging problems by combining the characteristics of image information distribution, the advantages of various image processing technologies, and Retinex theory. Furthermore, this method can effectively suppress halo artifacts and reduce image noise. Firstly, we decompose the input image into a base layer and a detail layer. Next, we convert the base layer image to Hue Saturation Value (HSV) color space and propose an improved Retinex model to handle the Value channel separately. Moreover, we propose an image gradient denoising method based on fast guided filtering for noise reduction and detail enhancement on the detail layer image. Finally, two-layer processing results are fused in RGB space to obtain the final enhanced image. The experiment shows that our proposed method comprehensively addresses various challenging issues in existing low-light enhanced images and performs well in suppressing halo artifacts and noise.
Time series forecasting has been an active research area, particularly in the context of Digital Twin (DT) systems. Despite the excellent results yielded by pre-trained models in Natural Language Processing (NLP) and Computer Vision (CV), only a few studies have researched pre-training strategies for time series forecasting networks within DT systems. Recent studies demonstrate that transfer learning across multiple time series datasets does not always provide promising results, making self-supervised pre-training directly on the downstream dataset a temporarily considered optimal solution. To the best of our knowledge, the only general pre-training task in the time series field is the Masked Autoencoder. However, this approach may lead to redundant representations for downstream forecasting within DT systems. Therefore, we propose three pre-training tasks specially designed for time series forecasting within DT frameworks: Inverse Forecasting (IF), Coarser Forecasting (CF), and Anomaly Forecasting (AF). These tasks are respectively designed to capture bidirectional dependency, reduce noise, and augment data, all crucial aspects in enhancing the predictive capabilities of DT systems. By integrating the three tasks, we obtain a composite pre-training task which generally improves the forecasting results of time series models within DT systems across multiple datasets. This work contributes to the ongoing efforts to improve the accuracy and efficiency of DT systems, paving the way for more robust and reliable digital representations of physical systems.
In this paper, the effect of topology structures on filter out-of-band (OoB) rejection is analyzed and the approximate expressions for OoB rejection of different structures are provided. Using the Mason model, the optimal synthesis topology of $\pi$ -type and L-type is applied for designing a Band39(1880 MHz-1920 MHz) filter to achieve good performance of low OoB rejection exceeding -70 dB as well as a return loss less than -15 dB according to the simulation results.
In recent years, collaborative software development has gradually become an emerging software development model. However, with the continuous development of crowdsourcing platforms, the problem of information overload has become increasingly serious, and it has become crucial to recommend suitable developers for tasks. Traditional recommendation methods face two major challenges: first, the text features of tasks and developers are highly concise and contain a large number of entities; second, tasks are one-time and explicit interactive data is sparse. To solve these challenges, this paper proposes a developer recommendation algorithm based on multi-relationship knowledge enhancement. We identify entities and contextual entities from text features and link the relationship between tasks and developers from a knowledge perspective. In addition, we treat developers' participation in registering, submitting, and winning tasks as different preferences, and assign different weights to each relationship. Finally, we enhance developers' feature representation using multi-relationship neighborhood aggregation. We conducted extensive experiments on a real Topcoder dataset, and the results show that our method is significantly better than the other four comparison methods in accuracy and MRR.
Mobile games are a popular form of entertainment that attract many users, but it is a difficult problem to find the right game from a large number of games. Mobile game community platforms such as TapTap provide player ratings and reviews functions to help users understand the quality and features of games, but there are a lot of fake reviews among them, which may lead to users making wrong choices. In this paper, we propose a Clustering-based Reliable Review Recommender Model (CRRRM) for this problem. The model comprehensively uses the text, the attributes of reviewers and reviews, filters out reliable reviews from unlabeled data through fuzzy C-means algorithm, and then makes recommendations. We collect more than 400 thousand mobile game reviews data on the TapTap platform for experiments. The results show that our model significantly outperforms the state-of-the-art baseline model in recommendation performance.
The array is a new structure that can effectively improve the performance of the sensor. Many scholars have studied the role of charge accumulation in the array, but there is no basic performance study for one unit, that is, two sensors in series. In this paper, a sensor with the same size made of piezoelectric material PZT and magnetostrictive material is designed. It is connected in series to study the bias magnetic field, resonance frequency, resonance peak, sensitivity and linearity. Through the test of the series structure, we found that compared with a single sensor, when the input excitation does not become, the bias magnetic field of the series inductor does not change, the resonance frequency does not change, and the resonance peak is the sum of a single sensor. The sensitivity and sensitivity have also been significantly enhanced. It provides an idea for the study of multi-unit array sensor performance.
This study describes a possibility of using magnetoelectric coupling effect to measure ultra-low-frequency weak AC magnetic fields. The research uses a composite structure composed of a PZT piezoelectric body with a size of $\mathbf{25} \mathbf{mm} \times \mathbf{2}\ \mathbf{mm} \times \mathbf{0.5}\ \mathbf{mm}$ and an appropriate size of magnetostrictive material Metglas as a sensitive element. A Metglas/PZT-5B magnetoelectric coupled sensor with a mechanical resonance frequency $\boldsymbol{f}_{\boldsymbol{r}}$ of 58 kHz was prepared. And a test system was set up to measure the weak AC magnetic field and display it in real time. This system uses the frequency up-conversion technique (FUCT) of ultra-low-frequency magnetic field, combined with hardware devices and software programs, for the magnetoelectric coupled sensor made by this test system, it can test the ultra-low-frequency weak AC magnetic field with a frequency as low as 0.1 nT @ 1 Hz.
System-in-package (SiP) is an advanced packaging technology that integrates multiple chips and components into a single package, creating a highly integrated system. Integrated Passive Devices (IPD) technology is commonly used to design resistors, inductors, capacitors, and other passive circuits, enabling the development of new products and solutions that was not possible with traditional board solutions. In recent years, IPD technology has become an important way to achieve SiP. This paper presents a practical implementation of the Differential Transformer Coupled Configuration Circuit using IPD techniques in a RF signal processing SiP, which minimizes the system board area and improves system performance. The paper demonstrates a typical ADC differential input and DAC differential output differential transformer coupled circuit IPD design. The Balun chip and related discrete RLC components are designed as one IPD chip, which can be integrated with the ADC and DAC through a silicon interposer or plastic substrate using Flip chip mode or wire bond in Advanced Package. The dual-channel ADC signal input coupled circuit IPD design meets the application requirements of bandwidth ranging from 1.8 GHz to 5 GHz and return loss and amplitude balance less than 2.5 dB and 3 dB, respectively. The double-input channel coupled circuit chip using fanout integration process can be implemented using RDLs in Advanced Packaging. The results offer a new way for coupled circuit design that significantly reduces the design complexity and area of the system PCB board.
Most target tracking is performed on visible video. This conventional RGB video lacks spectral information, leading to the tracking process being susceptible to interference from background clutter, illumination changes, etc. Hyperspectral images (HSIs) have the advantage of spectral integration, which can interpret the features of a target in both spatial and spectral dimensions. Therefore, HSIs with continuous spectral information have unique advantages in the field of target tracking. In this paper, a correlation filter-based HSIs tracking method is proposed. Feature maps are generated using tensor singular spectrum analysis (tensorS SA) feature and depth features to jointly exploit spatial contextual information and deep semantic information. Then a kernelized correlation filtering framework is used to generate feature responses on HSIs, and the largest region in the final output response is the center of the target in the current frame. Experiments conducted on hyperspectral images show that the proposed method can effectively utilize spatial-spectral information and outperforms the hyperspectral video tracking methods that partially use single features and the traditional color video tracking methods.
