This paper proposes a 256-bit speed-area-efficient hardware elliptic curve point-multiplication engine (ECPM-engine) in GF(p) over generic Weierstrass curves, which is optimized by a new speed-area-efficient radix-64 Montgomery modular multiplication (R64MMM) and a novel Montgomery ladder scheduling. The R64MMM calls one 129-bit adder and one (64x64+129)-bit multiply-accumulator (64-129-MAC) in parallel to make a trade-off between speed and area. The novel Montgomery ladder scheduling is used to improve the utilization of MAC in ECPM operation. In this ECPM-engine, both MAC utilization in R64MMM operations and R64MMM utilization in ECPM operations are close to 100%. The result shows that the proposed ECPM-engine consumes 72k gates when the clock frequency is 714 MHz with a 90 nm standard cell library, and it computes one 256-bit ECPM in 0.14 ms.
This paper presents a design of a gyro-TWT operating in the large-orbit electron beam mode, with the aim of reducing the working magnetic field while ensuring operational efficiency. The periodic dielectric-loaded structure is adopted as the high-frequency interaction circuit, which has not been reported in the application of large-orbit gyro-TWT in literature, while this structure has been successfully applied in the development of small-orbit devices. This paper conducts a comprehensive study and analysis of this structure and achieves stable operation in the Ka-band after the optimization of the tube. This tube works at second harmonic of electron frequency in the mode of large-orbit electron beam. The required magnetic field is only 5100 Gauss, which can be generated using electromagnetic coils instead of superconducting magnets. The operational parameters include voltage of 75 kV, current of 9A, and velocity spread of 3.5%. Under these conditions, the device presents stable operation, with -3 dB bandwidth of 4.3 GHz, and maximum output power of 165 kW. This result meets the expected requirements for magnetic field and operational efficiency, thus validating the feasibility of practical fabrication of large-orbit gyro-TWT with periodic dielectric-loaded structure.
Semantic communication has attracted significant attention as a key technology for emerging 6G communications. Though it has lots of potentials specially for high volume media communications, still there is no proper quality metric for modelling the semantic noise in semantic communications. This paper proposes an autoencoder based image quality metric to quantify the semantic noise. An autoencoder is initially trained with the reference image to generate the encoder decoder model and calculate its latent vector space. Once it is trained, a semantically generated/received image is inserted to the same autoencoder to create the corresponding latent vector space. Finally, both vector spaces are used to define the Euclidean space between two spaces to calculate the Mean Square Error between two vector spaces, which is used to measure the effectiveness of the semantically generated image. Results indicate that the proposed model has a correlation coefficient of 88% with the subjective quality assessment. Furthermore, the proposed model is tested as a metric to evaluate the image quality in conventional image coding. Results indicate that the proposed model can also be used to replace conventional image quality metrics such as PSNR,SSIM,MSSIM,UQI, VIFP, and SSC whereas these conventional metrics completely failed in semantic noise modelling.
This paper solves the problem of block sparse vector recovery using the block $\ell_1-\alpha\ell_q$- minimization model. Based on the block restricted isometry property (B-RIP) condition, we obtain exact block sparse vector recovery result. We also obtain the theoretical bound for the block $\ell_1-\alpha\ell_q$- minimization model when measurements are depraved by the noises.
In this letter, we report the design of a low-power event-based vision sensor (EVS) digital pixel in a 3D-stacked 45-nm backside illumination CMOS process. The design uses in-pixel analog-to-digital converters and memory to store digitized light intensities. By using a bit comparator to compare digital intensity values of the same pixel in the time domain, the high-fidelity event stream and intensity information from the same pixel can be generated simultaneously. The power consumption of the proposed digital EVS pixel can be as low as 10 nW/pixel. The EVS sensor can achieve a maximum event rate of 260 Meps.
The previous point cloud compression methods only consider reducing the amount of data. However, in applications such as autonomous driving, the compression methods not only require smooth transmission, but also improve the efficiency of downstream tasks. To this end, we propose a task-driven sampling network based on graph convolution to achieve point cloud compression and recovery. First, we present a task-driven downsampling network based on graph convolution to compress the point cloud. Then, we present an upsampling network based on graph convolution to enhance and recover the point cloud. In order to optimize the compressed point cloud for task, we add the task loss to loss function for end-to-end training. Experiments for point cloud classification task on ModelNet40 dataset show that the compressed point cloud obtained through our network can achieve higher classification accuracy compared to other similar methods, and the reconstructed point cloud can further improve classification accuracy.
Deforestation of the Amazon rainforest is approaching the worst in history. To protect against deforestation, it is necessary to accurately estimate the deforestation area. However, it is difficult to analyze large areas without direct human access. In addition, even if deforestation is estimated using satellite images, the presence of extensive cloud cover during the rainy season makes it challenging to obtain a clear view of the ground surface. In this paper, we propose a segmentation method based on deep learning and post-processing to predict the deforestation status in the Amazon rainforest area. To train and predict the deforestation area, we utilize a multi-modal satellite imagery dataset, including Sentinel-1, Sentinel-2, and Landsat 8. The proposed approach achieves the highest performance in the official CVPR MultiEarth Workshop 2023 challenge.
A 3D-printed X-band evanescent mode waveguide filter based on pixelization strategy is proposed in this letter. The uniqueness of the filter lies in the exploitation of the evanescent mode waveguide in conjunction with a 3D-printed dielectric pixelated structure. The crucial advantage of this approach is the fact that the dielectric pixelated structure located in a waveguide is not metal coated, so the fabrication process is easier. The proposed filter is manufactured using the fused deposition modeling 3D print technology. Its measured transmission coefficient in the passband is approximately -2 dB, and the reflection coefficient is below -19 dB.
A three-step discrete-time incremental analog-to-digital converter (IADC) combines zoom, IΔΣM and dual-mode SAR-assisted extended counting (EC). The IADC reuses the SAR ADC to reduce hardware cost by reconfiguring its DAC array to either a 2-bit quantizer of the core IΔΣM or a 5-bit EC ADC. Clocked at 4MHz with an OSR=99, the proposed IADC achieves SNDR and DR of 93dB and 98.5dB, respectively, in a BW of 20.2kHz.
To address significant disturbances and unavoidable measurement noises during aeroengine transient tests in the Intake Environmental Pressure Simulation System (IEPSS), a sliding mode active disturbance rejection controller (SMADRC) is designed. Meanwhile, a particle swarm optimization (PSO) algorithm is adopted for optimizing and tuning the parameters of the proposed controller. Then, experiments are conducted on IEPSS simulation platform to verify the proposed method. The results demonstrate that the proposed controller has better noise tolerance and robustness, and eventually enhances the system comprehensive control performance.
This paper presents a millimeter-wave Integrated Substrate Gap Waveguide (ISGW) filtering antenna with four controllable radiation nulls, two on each of its upper and lower stop bands. These radiation nulls can be adjusted by manipulating the dimensions of the stepped-impedance resonators (SIRs), complementary U-slots, and passive coplanar parasitic patches. This filtering antenna has the advantages of separately controllable radiation nulls and flexible adjustment of selectivity and gain curve roll-off. The simulation results demonstrate that the antenna operates at a center frequency of 25.4 GHz, with a relative bandwidth of 14.3% (23.76-27.04 GHz), and achieves a realizable average gain of 7.6 dBi.
The rapid development of Industrial Internet has brought new changes to the system architecture for industrial manufacturing. This paper proposes a system architecture applicable to small-scale personalization production lines, the key of which is a three-layer architecture with the Internet layer, data layer, and field control layer, in which the user's requirements for personalization product are transmitted from the Internet to the production line and the production equipment realizes the personalized production of products by motion control. This system architecture for production lines integrates Industrial Internet and smart manufacturing technologies, which reduces the complexity of the enterprise system architecture and increases the flexibility of the system compared to mass personalization production systems. The system architecture has been proven in a flexible production line for yogurt filling and can lay the foundation for the trans-formation of industrial manufacturing to a personalization production.
In the actual communication environment, the Specific emitter identification (SEI) task often encounters Zero-Shot Learning (ZSL) problems. In the ZSL scenarios, the irrelevant characteristics may weaken the inter-class aggregation and split features of the same category into different clusters, which exacerbates the misjudgment. In this paper, based on the global modeling ability of graph convolutional network (GCN), we design a Neighbourhood grAph NetwOrk (NANO) to improve this situation, which consists of feature extraction and a GCN-based transformation network. To train this network, we define a neighborhood graph (NG), a weighting strategy, and a novel NG loss. Finally, experiments on practical collected signals demonstrate that DNG-Net outperforms discrete value detection methods in SEI tasks.
The medical robot systems have the effect of reducing the work load and preventing accidents by supporting task that are difficult for inexperienced medical practitioners. Recently, research on automated blood collection and injection robotic system has been increasing, but work is still necessary such as replacing needle tips, setting tourniquet and fixing the patient's arm. In this letter, we propose an intelligent intravenous injection robot without having to see the medical workers face-to-face. We design a revolving system for automatic needle replacement and an arm holder for vessel fixation. Finally, we experiment positioning the needle at the target point by position control based on mathematical modeling and velocity kinematics.
Localization in GNSS-denied environments for Unmanned Aerial Vehicles (UAVs) has recently gained significant interest from the research community. Most of the research is focused primarily on visual localization. This paper, examines an algorithm which employs Angle of Departure (AoD) and UAVs payload sensor data for UAV localization. First the algorithm uses multiple AoDs from a single base station and a travel calculated by applying dead-reckoning on the UAVs Inertial Measurement Unit (IMU), to compute UAV location in two-dimensional (2D) coordinates. The 2D location estimate is then fed into a modified Extended Kalman Filter (EKF), which employs the estimate, IMU and barometer data to compute the three-dimensional (3D) coordinates for UAV. For the simulation, we applied Simulation-in-the-Loop (SITL) accompanied by Arducopter and MAVLink to simulate different trajectories and collect the required data for the algorithm. We validated our algorithm by comparing the EKF estimates with IMU dead-reckoned positions. Three simulations were performed, consisting of linear, zigzag and curved trajectories. We achieved a 90th percentile error of 2.5m and 4m for the x-coordinate and y-coordinate, respectively, on the zigzag and curved trajectories. Interestingly, the linear trajectory showed a larger localization error in its y-coordinate.
Nanoscale memristors open up new opportunities for the development of brain neural networks. Simple and precise memristors enhance the performance of various neural networks and operational circuits. In this letter, a three-terminal memristor is proposed, which makes the memristor more flexible and practical in circuit design and application through the introduction of a control port. Consid-ering that the resistance of a three-terminal memristor consists of three parts, i.e., metal region, low-resistance region, and high-resistance region, a three-segment piecewiselinear method is applied to fit these three regions. The model of this memristor is constructed through the derivation of the memristor formula and working principle. Candence simulations are conducted on the resultant circuit to verify its correctness.
This paper presents a voltage-mode direct time-of-flight (DToF) driver with high resolution auto-peak-power controller (APPC) and peak-current detector (PCD), utilizing on-chip oscilloscope (OCO) technology. The OCO supports both optical input for APPC and electrical input for PCD. The optical input detects laser diode (LD) peak power through photodiode (PD) first, and then feeds back to the boost converter to adjust laser diode supply voltage (LDVCC) to reach the target peak optical power. The electrical input detects the peak current of the LD to prevent the APPC from adapting to excessive output caused by the abnormal operation of the PD or LD. The X-coordinate of OCO has an 11-bit precision. It can detect pulse widths ranging from 500 ps to 10 ns. The test result shows that, the variation of optical power can be controlled within 2\% in the temperature range of 25-85 ℃ with APPC function.