Search

The application of SiC-based strain-relaxed buffers (SRB) technology in gate-all-around (GAA) pMOS nanosheet transistors (NS-FETs) fabrication has been systematically investigated. TCAD simulation results show that SiC SRB can effectively enhance the p-channel stress, up to 3.8Gpa has been achieved without S/D parasitic RC degradation. Furthermore, introducing a wide-bandgap SiC layer underneath NS-FET can help suppress the bottom parasitic transistor. The SiC SRB technology presents a integrated and streamlined approach for addressing the major performance bottlenecks of NS-FETs and is a potential solution for developing future NS-FET based high-performance and low-power logic applications.

Investigating neural mechanisms of anesthesia process and developing efficient anesthetized state detection methods are especially on high demand for clinical consciousness monitoring. Traditional anesthesia monitoring methods are not involved with the topological changes between electrodes covering the prefrontal-parietal cortices, by investigating electrocorticography (ECoG). To fill this gap, a framework based on the two-stream graph convolutional network (GCN) was proposed, i.e., one stream for extracting topological structure features, and the other one for extracting node features. The two-stream graph convolutional network includes GCN Model 1 and GCN Model 2. For GCN Model 1, brain connectivity networks were constructed by using phase lag index (PLI), representing different structure features. A common adjacency matrix was founded through the dual-graph method, the structure features were expressed on nodes. Therefore, the traditional spectral graph convolutional network can be directly applied on the graphs with changing topological structures. On the other hand, the average of the absolute signal amplitudes was calculated as node features, then a fully connected matrix was constructed as the adjacency matrix of these node features, as the input of GCN Model 2. This method learns features of both topological structure and nodes of the graph, and uses a dual-graph approach to enhance the focus on topological structure features. Based on the ECoG signals of monkeys, results show that this method which can distinguish awake state, moderate sedation and deep sedation achieved an accuracy of 92.75% in group-level experiments and mean accuracy of 93.50% in subject-level experiments. Our work verifies the excellence of the graph convolutional network in anesthesia monitoring, the high recognition accuracy also shows that the brain network may carry neurological markers associated with anesthesia.

The Reconfigurable Intelligent Surface (RIS) constitutes one of the prominent technologies for the next generation of wireless communications. It is envisioned to enhance the signal coverage in cases when the direct link of the communication is weak. Recently, beam training based on codebook selection is proposed to obtain the optimized phase configuration of the RIS. After that, the data is transmitted and received by using the classical coherent demodulation scheme (CDS). This training approach is able to avoid the large overhead required by the channel sounding process, and it also circumvents complex optimization problems. However, the beam training still requires the transmission of some reference signals to test the different phase configurations of the codebook, and the best codeword is chosen according to the measurement of the received energy of the reference signals. Then, the overhead due to the transmission of reference signals reduces the spectral efficiency. In this paper, a zero overhead beam training for RIS is proposed, relying on data transmission and reception based on non-CDS (NCDS). At the BS, the received differential data can also be used for the determination of the best beam for the RIS. Therefore, the efficiency of the system is significantly enhanced since reference signals are fully avoided. After choosing the best codebook, NCDS is still more suitable to transmit information for high mobility scenarios as compared to the classical CDS. Analytical expressions for the Signal-to-Interference and Noise Ratio (SINR) for the non-coherent RIS-empowered system are presented. Moreover, a detailed comparison between the NCDS and CDS in terms of efficiency and complexity is also given. The extensive computer simulation results verify the accuracy of the presented analysis and showcase that the proposed system outperforms the existing solutions.

The gravity recovery and climate experiment (GRACE) provides an unprecedented opportunity to detect the spatial and temporal variation of the terrestrial water storage (TWS) for regional to continental scales. However, the GRACE system's coarse temporal resolution (∼monthly) and data discontinuity missing perplexed the TWS research during the operation. In this article, the data assimilation (DA) method was employed to integrate the vertical deformation obtained from the small baseline subset (SBAS) InSAR processing into the NASA catchment land surface model (CLSM), which improved the estimation of the TWS. First, we used a one-dimensional ensemble Kalman filter for DA research to estimate the TWS in Dali Prefecture, southwestern China. Finally, we compared the estimated TWS with the GRACE-based TWS from December 2, 2018 to January 21, 2021. The unbiased root-mean-square of the open loop (OL; without DA) method and the SBAS-InSAR DA method are 61 mm and 30 mm in Dali Prefecture, respectively. Results revealed that the numerical difference between the estimated TWS and the GRACE TWS retrievals was significantly decreased by the SBAS-InSAR DA method than the OL method. In addition, the temporal resolution of the SBAS-InSAR DA-based TWS was improved to 12 days compared with GRACE-based TWS. Furthermore, we recovered the discontinuous deletion and blank of GRACE-based TWS from 2015 to 2018 by the SBAS-InSAR DA method.

The next generation Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2)/Advanced Topographic Laser Altimeter System (ATLAS) was launched in 2018 to provide large amounts of spaceborne laser altimetry data. The joint block adjustment with laser altimetry data and satellite remote sensing images can promote limited stereoscopic positioning accuracy without ground control points (GCPs). However, there are two problems in the joint block adjustment, that is, the reliable ATLAS laser altimetry points (LAPs) selection and the discrepancies of image-object points. To solve the above problems, a surface point cloud extraction algorithm based on wavelet reconstruction was proposed. Then, the LAPs selection constraint model was established, so the terrain influence was resolved. Finally, a block joint adjustment method has been proposed with remote sensing images and the reliable LAPs considering error of their plane coordinates, which could achieve the accurate corresponding image points of the reliable LAPs, remarkably decrease the discrepancies of image-object points. Fourteen pairs of ATLAS data, seven ZY3-02 stereo images and eleven Mapping Satellite-1 stereo images in Zhengzhou, China were collected to validate the method performance. Experiment results have shown that the consistency of the image-object points and the positioning accuracy of stereoscopic images without GCPs have been improved. Compared with the free net block adjustment, the height, east and north positioning accuracy of ZY3-02 and Mapping Satellite-1 stereo images have increased by 61%, 56%, 60% and 56%, 38%, 37%, respectively.

Digital data trails from disparate sources covering different aspects of student life are stored daily in most modern university campuses. However, it remains challenging to (i) combine these data to obtain a holistic view of a student, (ii) use these data to accurately predict academic performance, and (iii) use such predictions to promote positive student engagement with the university. To initially alleviate this problem, in this article, a model named Augmented Education (AugmentED) is proposed. In our study, (1) first, an experiment is conducted based on a real-world campus dataset of college students (N =156 ) that aggregates multisource behavioral data covering not only online and offline learning but also behaviors inside and outside of the classroom. Specifically, to gain in-depth insight into the features leading to excellent or poor performance, metrics measuring the linear and nonlinear behavioral changes (e.g., regularity and stability) of campus lifestyles are estimated; furthermore, features representing dynamic changes in temporal lifestyle patterns are extracted by the means of long short-term memory (LSTM). (2) Second, machine learning-based classification algorithms are developed to predict academic performance. (3) Finally, visualized feedback enabling students (especially at-risk students) to potentially optimize their interactions with the university and achieve a study-life balance is designed. The experiments show that the AugmentED model can predict students' academic performance with high accuracy.

Pilot symbol assisted modulation (PSAM) is widely used to obtain the channel state information (CSI) needed for coherent demodulation. It allows the density of pilot symbols to be dynamically chosen depending on the channel conditions. However, the insertion of pilots reduces the spectral efficiency, more severely when the channel is highly time-variant and/or frequency-selective. In these cases a significant amount of pilots is required to properly track the channel variations in both time and frequency dimensions. Alternatively, non-coherent demodulation does not require any CSI for the demodulation independently of the channel conditions. For the particular case of up-link (UL) based on massive single input - multiple output (SIMO) combined with orthogonal frequency division multiplexing (OFDM), we propose to replace the traditional reference signals of PSAM by a new differentially-encoded data stream that can be non-coherently detected. The latter can be demodulated without the knowledge of the CSI and subsequently used for the channel estimation. We denote our proposal as hybrid demodulation scheme (HDS) because it exploits both the benefits of a coherent demodulation scheme (CDS) and a non-coherent demodulation scheme (NCDS) to increase the spectral efficiency. The mean squared error (MSE) of the channel estimation, bit error rate (BER), achieved throughput and complexity are analyzed to highlight the benefits of this differential data-aided channel estimation as compared to other approaches. We show that the channel estimation is almost as good as PSAM, while the BER performance and throughput are improved for different channel conditions with a very small complexity increase.

Edge computing provides off-load computing and application services close to end-users, greatly reducing cloud pressure and communication overhead. However, wireless edge networks still face the risk of network attacks. To ensure the security of wireless edge networks, we present Federated Learning-based Attention Gated Recurrent Unit (FedAGRU), an intrusion detection algorithm for wireless edge networks. FedAGRU differs from current centralized learning methods by updating universal learning models rather than directly sharing raw data among edge devices and a central server. We also apply the attention mechanism to increase the weight of important devices, by avoiding the upload of unimportant updates to the server, FedAGRU can greatly reduce communication overhead while ensuring learning convergence. Our experimental results show that, compared with other centralized learning algorithms, FedAGRU improves detection accuracy by approximately 8%. In addition, FedAGRU's communication cost is 70% less than other federated learning algorithms, and it exhibits strong robustness against poisoning attacks.

Bearings-only tracking is a continuous-discrete system, whose state of motion is in the continuous-time domain and the measurement is in the discrete-time domain. The unpredicted approximation errors are inevitable due to integration, discretization, and linearization of continuous model in many filtering methods. The adaptive covariance feedback framework is proposed for solving this kind of problem, in which the posterior covariance sequence is proved theoretically to be useful for prior covariance updating. In this framework, the covariance feedback framework is integrated with the continuous-discrete cubature Kalman filtering, and Chebyshev distance is applied to judge the proper condition for the start-up of the feedback channel. The numerical results illustrate the proposed method’s superior performance in accuracy and computational efficiency.

Superimposed training (ST) is a semiblind channel estimation technique, proposed for orthogonal frequency division multiplexing (OFDM), where training sequences are added to data symbols, avoiding the use of dedicated pilot-subcarriers, and increasing the available bandwidth compared with pilot symbol assisted modulation (PSAM). Filter bank multicarrier offset quadrature amplitude modulation (FBMC-OQAM) is a promising waveform technique considered to replace the OFDM, which takes advantage of well-designed filters to avoid the use of cyclic prefix and reduce the out-band-emissions. In this paper, we provide the expressions of the average channel capacity of the FBMC-OQAM combined with either PSAM or ST schemes, considering imperfect channel estimation and the presence of the pilot sequences. In order to compute the capacity expression of our proposal, ST-FBMC-OQAM, we analyze the channel estimation error and its variance. The average channel capacity is deduced considering the noise, data interference from ST, and the intrinsic self-interference of the FBMC-OQAM. Additionally, to maximize the average channel capacity, the optimal value of data power allocation is also obtained. The simulation results confirm the validity of the capacity analysis and demonstrate the superiority of the ST-FBMC-OQAM over existing proposals.

An accurate recognition of a dimensional variation pattern is very important for producing high-quality body-in-white (BIW). The wide application of optical coordination measurement machines (OCMM) in vehicle factory provided massive online dimensional data for the variation pattern recognition. However, the massive serially correlated or autocorrelated and 100% measurement data generated from the OCMM challenge the traditional statistical process control (SPC) technology and the common variation recognition approaches. This paper presents a novel deep-learning method, long short-term memory neural network (LSTM NN), to recognize the variation pattern of the BIW OCMM online measurement data. A comparative study between the backpropagation neural network (BP NN) and the LSTM NN was implemented, and the practicability of the proposed intelligent method was demonstrated by a case study. With the efficient use of time series information, the LSTM NN has a good performance in variation patterns' recognition and high practicability in improving the quality of the BIW.

At present, deep learning has been widely adopted in medical image processing. However, the current deep neural networks depend on a large number of labeled training data, but medical images segmentation tasks often suffer from the problem of small quantity of labeled data because labeling medical images is a very expensive and time-consuming task. In order to overcome this difficulty, this paper proposes a new image augmentation strategy based on statistical shape model and three-dimensional thin plate spline, which can generate many simulated images from a small number of real images. Firstly, the shape information of the real labeled images is modeled with the statistical shape model, and a series of simulated shapes are generated by sampling from this model. Secondly, the simulated shapes are filled with texture using three-dimensional thin plate spline to generate the simulated images. Finally, the simulated images and the real images are used together for training deep neural networks. The proposed framework is a general data augmentation method that can be used in any anatomical structure segmentation tasks with any deep neural network architecture. We used two different datasets, including prostate MRI dataset and liver CT dataset, and used two different deep network structures, including multi-scale 3D Convolutional Neural Networks (multi-scale 3D CNN) and U-net. The experimental results showed that the proposed data augmentation strategy can improve the accuracy of existing segmentation algorithms based on deep neural networks.

In order to improve filtering accuracy and restrain the degradation of filtering performance caused by the heavy-tailed process and measurement noises in multi-target tracking, this paper proposes a robust Student's t mixture probability hypothesis density (PHD) filter. In the proposed method, a Student's t mixture is implemented to the PHD filter, which recursively propagates the intensity as a mixture of Student's t components in PHD filtering framework. Furthermore, with the advantage of a designed judging and re-weighting mechanism, an M-estimation-based dual-gating strategy is designed for the Student's t mixture implementation to suppress the negative effect of the heavy-tailed noises. Our proposed approach not only utilizes the Student's t distribution to match the real heavy-tailed non-Gaussian noise well but also enhances the robustness of the Student's t mixture-based approach via the designed dual-gating strategy. The simulation results verify that the proposed algorithm can keep good filtering accuracy in the presence of the process and measurement outliers simultaneously.

In the next-generation cognitive radio networks, numerous secondary users will share the spectrum resource with the primary users. As it may not be possible to support all the communication rate requirements, there are many supporting sets for the secondary users as long as the communication rates of the primary users are guaranteed. In this paper, we study the maximum feasible set problem to access as many secondary users as possible, under the constraints of power budgets and communication rates in cognitive radio networks. In this interesting issue, the existing literature generally removes a subset of the secondary users so that the remaining users achieve the thresholds with communication rates and power budgets. However, the removal algorithms cause more interference when there are plenty of unsupported secondary users. We leverage the spectral radius of the network characteristic matrix as the admission price to access the new secondary user. Then, we design a hybrid access control algorithm to reduce the interference time and approximate the maximum network capacity. Moreover, different supported sets produce the different energy efficiency, even having the same network capacity, while all users require the high communication rates. Numerical results demonstrate that our algorithms provide the decent energy efficiency under the communication rate constraints.

We report a spectrally and temporally cascaded dual-soliton Stokes excitation source for broadband coherent anti-Stokes Raman spectroscopy (CARS) based on a single Yb-doped fiber laser. The strong birefringence of a polarization-maintaining photonic crystal fiber has the potential to simultaneously generate two solitons along different eigenpolarization axes, allowing for an extra degree of freedom in tuning the properties of the supercontinuum. With well-separated spectral regions of this dual-soliton, the CARS spectral coverage can be greatly extended. Taking advantage of the dispersion characteristic of the two orthogonal polarized modes and applying identical linear frequency chirp on both soliton Stokes pulses further allows that spectral CARS information can be encoded and acquired in time. The chirped pump pulse, in conjunction with dual-soliton Stokes pulses, can potentially yield gapless coverage of Raman transitions from 800 to 1800 cm-1 with a spectral resolution of 15 cm-1. Broadband spectral coverage capability is theoretically verified with simulations and then experimentally demonstrated by CARS spectroscopy with varied polymer materials. We also demonstrate hyperspectral Raman microscopy with molecular contrast of lipid droplets. An all-fiber-based broadband excitation source may greatly simplify CARS implementation and extend the qualitative and even quantitative chemical analysis of CARS microspectroscopy.