检索结果-南通市图书馆

science China(information sciences) 2024年第2期67卷 267-278页

作者： Guofang YU Renrong LIANG Haiming ZHAO Lei XIAO Jie CUI Yue ZHAO Wenpu CUI Jing WANG Jun XU Jun FU Tianling REN School of Integrated Circuits Tsinghua University beijing national research center for information science technology Tsinghua University

Surface acoustic wave(SAW) resonators with an ultrahigh Q-factor are designed and fabricated on silicon-based gallium nitride(GaN/Si). The temperature-dependent performance is characterized over a wide range, from 10 to 500 K. Finite element analysis is employed to guide the design of the SAW resonator from indications of the Rayleigh mode and weak propagation direction dependence of SAW in the c-plane of GaN/Si. The SAW resonator with 100 pairs of interdigital transducers(IDT), 100 pairs of grating reflectors(GR) for each side, aperture size of 80 μm, metallization ratio of 0.5, and electrode width of 500 nm resonates at 1.9133 GHz accordingly with an ultrahigh Q-factor of 7622 at room temperature, which contributes the fr× Qr, up to 14.583×1012Hz. A resonator operating over 10 to 500 K indicates an approximately linear decreasing temperature dependence above 280 K while being approximately constant below 40 K. The fitting to resonator characteristics using the modified Butterworth Van Dyke(m BVD) model reveals a reduction in both the electrode and mechanical losses while worsening the dielectric loss with cooling down.

关键词： GaN/Si SAW resonator ultrahigh Q-factor temperature-dependent performance mBVD losses

Architecture-circuit-technology co-optimization for resistive random access memory-based computation-in-memory chips

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science China(information sciences) 2023年第10期66卷 152-161页

作者： Yuyi LIU Bin GAO Jianshi TANG Huaqiang WU He QIAN School of Integrated Circuits Beijing National Research Center for Information Science and Technology(BNRist) Tsinghua University

Computation-in-memory(CIM) chips offer an energy-efficient approach to artificial intelligence computing workloads. Resistive random-access memory(RRAM)-based CIM chips have proven to be a promising solution for overcoming the von Neumann bottleneck. In this paper, we review our recent studies on the architecture-circuit-technology co-optimization of scalable CIM chips and related hardware demonstrations. To further minimize data movements between memory and computing units, architecture optimization methods have been introduced. Then, we propose a device-architecture-algorithm co-design simulator to provide guidelines for designing CIM systems. A physics-based compact RRAM model and an array-level analog computing model were embedded in the simulator. In addition, a CIM compiler was proposed to optimize the on-chip dataflow. Finally, research perspectives are proposed for future development.

关键词： resistive random-access memory computation-in-memory compact model device-architecture-algorithm co-design compiler

High-speed optoelectronic devices

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science China(information sciences) 2023年第5期66卷 5-27页

作者： Yi LUO Changzheng SUN Bing XIONG Jian WANG Zhibiao HAO Yanjun HAN Hongtao LI Lai WANG beijing national research center for information science and technology (BNRist) Department of Electronic Engineering Tsinghua University

High-speed optoelectronic devices are key components of modern fiber communication systems,and the backbone of information technology. In this paper, we present our work on high-speed devices over the past decades, including high-performance semiconductor lasers and integrated light sources, wideband electrooptic modulators, high saturation power photodetectors, and their applications in both fiber communications and microwave photonics. Gain-coupled distributed feedback(DFB) lasers were fabricated with high singlemode yield. Electroabsorption modulated lasers(EMLs) were demonstrated based on identical epitaxial layer integration scheme, including single channel 40 Gb/s EMLs, and 4 × 25 and 4 × 56 Gb/s EML arrays. Both In P and thin-film lithium niobate based low half-wave voltage electro-optic modulators with modulation bandwidth over 40 GHz were demonstrated. Back-illuminated modified uni-traveling-carrier photodetectors(MUTC-PDs) with high saturation power and bandwidth exceeding 100 GHz were developed. Applications of high-speed optoelectronic devices in fiber-optic links and sub-THz wave generation are presented.

关键词： optoelectronic devices semiconductor lasers optical modulators photodetectors integrated photonic circuits

Near-field wideband channel estimation for extremely large-scale MIMO

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science China(information sciences) 2023年第7期66卷 283-296页

作者： Mingyao CUI Linglong DAI Department of Electronic Engineering Tsinghua University Beijing National Research Center for Information Science and Technology(BNRist)

Extremely large-scale multiple-input-multiple-output(XL-MIMO) at millimeter-wave(mm Wave)and terahertz(THz) bands plays an important role in supporting extreme high beamforming gain as well as ultra-wideband spectrum resources. Unfortunately, accurate wideband XL-MIMO channel estimation suffers from a new challenge called the near-field beam split effect. Prior studies either neglect the accurate nearfield channel model or fail to exploit the beam split effect, resulting in poor channel estimation accuracy for wideband XL-MIMO. To tackle this problem, this paper proposes a bilinear pattern detection(BPD)-based approach to accurately recover the wideband XL-MIMO channel. Specifically, by analyzing the characteristics of near-field wideband channels, we first reveal the bilinear pattern of the near-field beam split effect, which implies that the sparse support set of near-field channels in both the angle and the distance domains can be regarded as a linear function against frequency. Then, inspired by the classical simultaneously orthogonal matching pursuit technique, we use the bilinear pattern to estimate the angle-of-arrival(Ao A) and distance parameters of each near-field path component at all frequencies. In this way, the entire wideband XL-MIMO channel can be recovered by compressed sensing algorithms. Moreover, we provide the computational complexity of the proposed algorithm compared with existing algorithms. Finally, simulation results demonstrate that our scheme can achieve the accurate estimation of the near-field wideband XL-MIMO channel in the presence of near-field beam split effect.

关键词： XL-MIMO wideband near-field beam split channel estimation

Towards real-time ML-based DDoS detection via cost-efficient window-based feature extraction

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science China(information sciences) 2023年第5期66卷 114-129页

作者： Haibin LI Yi ZHAO Wenbing YAO Ke XU Qi LI Department of Computer science and technology Tsinghua University beijing national research center for information science and technology (BNRist) Institute for Network sciences and Cyberspace Tsinghua University

Distributed denial of service(DDoS) detection is still an open and challenging problem. In particular, sophisticated attacks, e.g., attacks that disguise attack packets as benign traffic always appear, which can easily evade traditional signature-based methods. Due to the low requirements for computing resources compared to deep learning, many machine learning(ML)-based methods have been realistically deployed to address this issue. However, most existing ML-based DDo S detection methods are highly dependent on the features extracted from each flow, which incur remarkable detection delay and computation overhead. This article investigates the limitations of typical ML-based DDo S detection methods caused by the extraction of flow-level features. Moreover, we develop a cost-efficient window-based method that extracts features from a fixed number of packets periodically, instead of per flow, aiming to reduce the detection delay and computation overhead. The newly proposed window-based method has the advantages of well-controlled overhead and wide support of common routers due to its simplicity and high efficiency by design. Through extensive experiments on real datasets, we evaluate the performance of flow-based and window-based methods.The experimental results demonstrate that our proposed window-based method can significantly reduce the detection delay and computation overhead while ensuring detection accuracy.

关键词： DDoS attack machine learning feature extraction detection delay cost-efficiency

Array-level boosting method with spatial extended allocation to improve the accuracy of memristor based computing-in-memory chips

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

science China(information sciences) 2021年第6期64卷 80-88页

作者： Wenqiang ZHANG Bin GAO Peng YAO Jianshi TANG He QIAN Huaqiang WU Institute of Microelectronics Beijing National Research Center for Information Science and Technology (BNRist)Tsinghua University Institute of Microelectronics Beijing National Research Center for Information Science and Technology (BNRist)Tsinghua University

Memristor based computing-in-memory chips have shown the potentials to accelerate deep neural networks with high energy efficiency. Due to the inherent filament-based conductive mechanism of the memristor, the reading and writing noises are hard to eliminate. Besides, the precision of the large-scale memristor array is still limited. However, when the noise of the memristor is large, the existing training methods to reduce the accuracy loss of memristor based computing-in-memory chips will face challenges.Hence, we proposed the array-level boosting method with spatial extended allocation to reduce the accuracy loss induced by the limited precision and large noises. To optimize the spatial allocation number of each layer in the neural network, the greedy spatial extended allocation algorithm is also proposed. The image processing and classification tasks are demonstrated based on fabricated 32×128 memristor arrays to valid the performance of the proposed method. The chip-in-loop results show that the recovered accuracy of ResNet-34 on CIFAR-10 with array-level boosting method is 92.3%, which is closed to software-based accuracy of 93.2%.

关键词： memristor computing-in-memory array-level boosting neuromorphic computing RRAM

Eavesdropping Area for Evaluating the Security of Wireless Communications

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

China Communications 2022年第3期19卷 145-157页

作者： Xiaofeng Zhong Chenchen Fan Shidong Zhou Tsinghua University Beijing National Research Center for Information Science and Technology Department of Electronic Engineering Tsinghua UniversityBeijing 100084China

Compared with wired communication,the wireless communication link is more vulnerable to be attacked or eavesdropped because of its broadcast nature.To prevent eavesdropping,many researches on transmission techniques or cryptographic methods are carried out.This paper proposes a new index parameter named as eavesdropping area,to evaluate the anti-eavesdropping performance of wireless system.Given the locations of legitimate transmitter and receiver,eavesdropping area index describes the total area of eavesdropping regions where messages can be wiretapped in the whole evaluating region.This paper gives detailed explanations about its concept and deduces mathematical formulas about performance curves based on region classification.Corresponding key system parameters are analyzed,including the characteristics of eavesdropping region,transmitted beam pattern,beam direction,receiver sensitivity,eavesdropping sensitivity,path loss exponent and so on.The proposed index can give an insight on the confirmation of high-risk eavesdropping region and formulating optimal transmitting scheme for the confidential communications to decrease the eavesdropping probability.

关键词： eavesdropping area wireless communication anti-eavesdropping performance transmitting scheme

维普期刊数据库

Hybrid Satellite-UAV-Terrestrial Maritime Networks:Network Selection for Users on A Vessel Optimized with Transmit Power and UAV Position

在线全文

学校读者我要写书评

暂无评论

China Communications 2022年第9期19卷 37-46页

作者： Xiangling Li Wenjing Shi School of information science and technology Beijing University of Chemical TechnologyBeijing 100029China beijing national research center for information science and technology Tsinghua UniversityBeijing 100084China

The hybrid satellite-UAV-terrestrial maritime networks have shown great promise for broadband coverage at sea.The existing works focused on vessels collaboratively served by UAV-enabled aerial base station(ABSs)and terrestrial base stations(TBSs)deployed along the coast,and proved that data rate could be improved by optimizing transmit power and ABS’s position.In practice,users on a vessel can be collaboratively served by an ABS and a vesselenabled base station(VBS)in different networks.In this case,how to select the network for users on a vessel is still an open issue.In this paper,a TBS and a satellite respectively provide wireless backhaul for the ABS and the VBS.The network selection is jointly optimized with transmit power of ABS and VBS,and ABS’s position for improving data rate of all users.We solve it by finding candidates for network selection and iteratively solving transmit power and ABS’s position for each candidate.Simulation results demonstrate that data rate can be improved by collaborative coverage for users on a vessel.

关键词： maritime communications satellite-UAV-terrestrial network network selection power allocation

维普期刊数据库

LiFE:Deep Exploration via Linear-Feature Bonus in Continuous Control

在线全文

学校读者我要写书评

暂无评论

tsinghua science and technology 2023年第1期28卷 155-166页

作者： Jiantao Qiu Yu Wang Department of Electronic Engineering and beijing national research center for information science and technology(BNRist) Tsinghua UniversityBeijing 100084China

Reinforcement Learning(RL)algorithms work well with well-defined rewards,but they fail with sparse/deceptive rewards and require additional exploration strategies.This work introduces a deep exploration method based on the Upper Confidence Bound(UCB)bonus.The proposed method can be plugged into actor-critic algorithms that use deep neural networks as a critic.Based on the conclusion of the regret bound under the linear Markov decision process approximation,we use the feature matrix to calculate the UCB bonus for deep exploration.The proposed method is equivalent to the count-based exploration method in special cases and is suitable for general situations.Our method uses the last d-dimensional feature vector in the critic network and is easy to deploy.We design a simple task,“swim”,to demonstrate the principle of the proposed method to achieve exploration in sparse/deceptive reward environments.Then,we perform an empirical evaluation on sparse/deceptive reward version gym environments and Ackermann robot control tasks.The evaluation results verify that the proposed algorithm can perform effective deep explorations in sparse/deceptive reward tasks.

关键词： Reinforcement Learning(RL) Neural Network(NN) Upper Confidence Bound(UCB)

Recent Progress and Applications of HfO_(2)-Based Ferroelectric Memory

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

tsinghua science and technology 2023年第2期28卷 221-229页

作者： Xiao Liu Xiangshun Geng Houfang Liu Minghao Shao Ruiting Zhao Yi Yang Tian-Ling Ren beijing national research center for information science and technology(BNRist) School of Integrated CircuitsTsinghua UniversityBeijing 100084China

The discovery of ferroelectricity in hafnium oxide (HfO_(2)) based thin films in 2011 renewed the interest inferroelectrics. These new ferroelectrics possess completely different crystal morphology with conventional perovskiteferroelectrics, and present more robust ferroelectric properties upon aggressive scaling and compatibility withstandard integrated circuit fabrication processes. In this article, we give a brief introduction to the conventionalferroelectric memories, then review the basic properties, recent progress, and memory applications of theseHfO_(2)-based ferroelectrics.

关键词： ferroelectric memory non-volatile memory hafnium oxide fluorite structure material