Fudan University

Wetlands are rich in biodiversity, provide habitats for many wildlife species, and play a vital role in the transmission of bird-borne infectious diseases (e.g., highly pathogenic avian influenza). However, wetlands worldwide have been degraded or even disappeared due to natural and anthropogenic activities over the past two centuries. At present, major data products of wetlands have large uncertainties, low to moderate accuracies, and lack regular updates. Therefore, accurate and updated wetlands maps are needed for the sustainable management and conservation of wetlands. Here, we consider the remote sensing capability and define wetland types in terms of plant growth form (tree, shrub, grass), life cycle (perennial, annual), leaf seasonality (evergreen, deciduous), and canopy type (open, closed). We identify unique and stable features of individual wetland types and develop knowledge-based algorithms to map them in Northeast China at 10 m spatial resolution by using microwave (PALSAR-2, Sentinel-1), optical (Landsat (ETM+/OLI), Sentinel-2), and thermal (MODIS land surface temperature, LST) imagery in 2020. The resultant wetland map has a high overall accuracy of >95%. There were a total 154,254 km 2 of wetlands in Northeast China in 2020, which included 27,219 km 2 of seasonal open-canopy marsh, 69,158 km 2 of yearlong closed-canopy marsh, and 57,878 km 2 of deciduous forest swamp. Our results demonstrate the potential of knowledge-based algorithms and integrated multi-source image data for wetlands mapping and monitoring, which could provide improved data for the planning of wetland conservation and restoration.

The tin (IV) oxide (SnO2) electron transport layer (ETL) has been widely employed to fabricate high-performance perovskite solar cells (PSCs). It has been reported that carbon quantum dots (CQDs) can be used to enhance electron mobility of SnO2. However, an in-depth understanding of the driving force in this process is still lacking. Here, a high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) is employed, for the first time, to reveal the SnO2 crystal face changes with one new type of CQD doping. Synchrotron-based grazing incidence wide-angle X-ray scattering (GIWAXS) can penetrate the flexible substrate to detect the buried region of the perovskite layer, showing the crystallinity and phase purity of the perovskite are significantly improved with CQD-modified SnO2. The flexible n-i-p PSCs delivers a power conversion efficiency (PCE) up to 23.57% (22.75%, certificated), which is one of the highest values for single-junction n-i-p flexible PSCs. The corresponding n-i-p flexible modules achieve a PCE of 17.79% with aperture area ~ 24 cm². Furthermore, the flexible PSCs show excellent stability, preserving ≈95% of their initial efficiency after 1200 h under 40% relative humidity and 1-sun light irradiation at 25 °C, and maintained > 90% of initial efficiency after 2500 bending cycles at a bending radius of 6 mm.

Positioning is an essential service for various applications and is expected to be integrated with existing communication infrastructures in 5G and 6G. Though current Wi-Fi and cellular base stations (BSs) can be used to support this integration, the resulting precision is unsatisfactory due to the lack of precise control of the wireless signals. Recently, BSs adopting reconfigurable holographic surfaces (RHSs) have been advocated for positioning as RHSs’ large number of antenna elements enable generation of arbitrary and highly-focused signal beam patterns. However, existing designs face two major challenges: i) RHSs only have limited operating bandwidth, and ii) the positioning methods cannot adapt to the diverse environments encountered in practice. To overcome these challenges, we present HoloFed, a system providing high-precision environment-adaptive user positioning services by exploiting multi-band (MB)-RHS and federated learning (FL). For improving the positioning performance, a lower bound on the error variance is obtained and utilized for guiding MB-RHS’s digital and analog beamforming design. For better adaptability while preserving privacy, an FL framework is proposed for users to collaboratively train a position estimator, where we exploit the transfer learning technique to handle the lack of position labels of the users. Moreover, a scheduling algorithm for the BS to select which users train the position estimator is designed, jointly considering the convergence and efficiency of FL. Our performance evaluation based on simulations confirms that HoloFed achieves a 57%lower positioning error variance compared to a beam-scanning baseline and can effectively adapt to diverse environments.

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, and the Y-coordinate also has a 11-bit precision. It can detect pulse widths ranging from 500 ps to 10 ns. The chip also integrates multiple power protection circuits for the boost converter. The test result shows that, the variation of optical power can be controlled within 2% in the temperature range of 25-85 °C with APPC function.

The ubiquity of edge devices has led to a growing amount of unlabeled data produced at the edge. Deep learning models deployed on edge devices are required to learn from these unlabeled data to continuously improve accuracy. Self-supervised representation learning has achieved promising performances using centralized unlabeled data. However, the increasing awareness of privacy protection limits centralizing the distributed unlabeled image data on edge devices. While federated learning has been widely adopted to enable distributed machine learning with privacy preservation, without a data selection method to efficiently select streaming data, the traditional federated learning framework fails to handle these huge amounts of decentralized unlabeled data with limited storage resources on edge. To address these challenges, we propose a Self-supervised On-device Federated learning framework with coreset selection, which we call SOFed, to automatically select a coreset that consists of the most representative samples into the replay buffer on each device. It preserves data privacy as each client does not share raw data while learning good visual representations. Experiments demonstrate the effectiveness and significance of the proposed method in visual representation learning.

The coexistence needs of sensing and communication in millimeter-wave (mmW) bands have urgently driven the seamless integration of sensing and communication in the upcoming mmW era. However, the time-frequency competition between the two functions makes it difficult to accommodate both high sensing resolution and large communication capacity. In this paper, we have designed a W-band fiber-wireless link with the integrated sensing and communication functions enabled by electromagnetic polarization multiplexing. The ultra-wideband fiber-wireless link in W band is enabled by the asymmetrical single-sideband modulation along with the optical heterodyne up-conversion. The electromagnetic polarization multiplexing allocates the sensing and communication functions on two orthogonal electromagnetic polarizations, respectively. Thus, all time-frequency resources of the fiber-wireless link can simultaneously serve these two functions without any resource competition, contributing to an ultra-high spatial resolution and an ultra-large data capacity at the same time. Our experimental results show the spatial resolution of up to 15 mm and data rate as high as 92 Gbit/s were simultaneously realized in W band after delivering over a 10.8-m wireless distance. The overall improvement of both the sensing and communication performance, to the best of our knowledge, led to a record capacity-resolution quotient of 61.333 Gbit/s/cm. In addition, we have qualitatively investigated the integrated sensing and communication fiber-wireless link, in terms of the carrier frequency, system bandwidth, multi-mmW access, and electromagnetic polarization crosstalk.

We propose a new technology and design platform using combined common gate N/PFET CFET as basic element for CMOS circuit applications. Two CFET unit structures, namely common gate (CG) and N-gate (NG), are identified to form base design elements. Through the two units, all circuit logic functions in standard cell library and SRAM can be realized without significant process complication. A Multi-gradient Neural Network (MNN) based SPICE compact modeling methodology is developed for these CFET units. As an example, MNN model generation is illustrated for CG with the output matching well with the TCAD data within and beyond the range of model extraction. Circuit simulations are exercised using the MNN models and demonstrated successfully the expected circuit functionalities. As the CFET technology be adopted as mainstream in future, this novel design framework proposed would enable efficient logic design.

Unmanned Aerial Vehicles (UAVs)-assisted Multi-access Edge Computing (MEC) has emerged as a promising solution in B5G/6G networks. The high flexibility and seamless connectivity of UAVs make them well-suited for providing enhanced communications coverage and efficient computing support. Particularly in situations where ground facilities may be compromised or communication is unreliable. In this paper, we study joint dynamic service switching and resource allocation for multiple UAVs in MEC network. We consider the heterogeneity of tasks and UAVs and model the dynamic service process of UAVs as a sequential decision problem based on the Markovian decision process. To enable dynamic and intelligent UAV service, we first propose a centralized dynamic service algorithm DDBC based on deep reinforcement learning. However, given the training difficulties of the centralized algorithm, we propose a more promising distributed learning algorithm FLBF, which combines federated learning. We conduct extensive simulations to evaluate the effectiveness and advantages of the proposed algorithms. Our results show that both DDBC and FLBF can improve the model convergence speed by 50%, and reduce the system cost by 12.30% to 35.72% compared to comparative algorithms. Furthermore, simulations indicate that FLBF is well-suited for training models in UAV-assisted MEC networks.

The operational amplifier is a key building block in analog systems. However, the design process of the operational amplifier is time-consuming and heavily depends on engineers’ experiences. This paper presents OPAMP-Generator, an analog operational amplifier generator, which automates the full design flow from user-defined specifications to GDSII layout without human intervention. OPAMP-Generator includes behaviorallevel topology optimization, efficient sizing algorithm based on the classical gm/Id design methodology, and automated layout generation. The behavioral-level description of the opamp is represented by the directed acyclic graph (DAG) and a customized variational graph autoencoder is proposed to embed the discrete graph representation into a low-dimensional continuous space. The topology of the opamp can thus be optimized in the latent space, which greatly improves the optimization efficiency. The sizing algorithm based on gm/Id methodology can guarantee the quality of transistor-level circuit implementation. The constraints of the layouts can be naturally derived from the topology level, which facilities the automatic generation of layouts. Experimental results demonstrate that our proposed method can efficiently synthesize operational amplifiers with competitive performances compared to manual designs.

An ear-bar TWS earphone antenna with improved ear-to-ear path gain (PG <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E2E</sub> ) and efficiency is presented. The proposed antenna consists of an inverted-L antenna and a parasitic metal. By applying the parasitic metal, the vertical electric-field (E-field) around the head is enhanced and the E-field into the human head is weakened. Due to the stronger vertical E-field around the head, the proposed antenna achieves enhanced PG <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E2E</sub> . Also, for the weak E-field in the head, the improved antenna efficiency is realized. Two prototypes without and with the parasitic metal are fabricated and measured. The measured -6dB bandwidths of the two prototypes can cover the ISM band. With the help of the parasitic metal, the measured average efficiencies on the head within the 2400-2484 MHz band are improved from 19.1% to 30.9%. In the meanwhile, the measured PG <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E2E</sub> on the head is improved from -56 dB to -49.7 dB. The proposed ear-bar TWS earphone antenna with enhanced PG <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E2E</sub> and efficiency is promising for practical applications.

This article presents an 8-way serial power combined transformer-based quadrature digital power amplifier (DPA) with in-phase/quadrature (IQ)-reuse and Doherty techniques for high output power and high efficiency. An 8-way serial-combining transformer (SCT) power combiner is employed to achieve 4.1 W (36.1 dBm) peak output power. IQ-reuse and transformer-based Doherty techniques are adopted to enhance the efficiency at 3 and 6 dB power back-offs (PBOs). Implemented in 28 nm CMOS and powered by 1.1 and 2.2 V supply voltages, this quadrature DPA achieves $>$ 30% peak power-added efficiency (PAE) across 2.7–3.2 GHz and its 1 dB radio frequency (RF) bandwidth ranges from 2.6 to 3.2 GHz. Besides, it obtains the PAE of 25.3% and 22.6% at 6 and 3 dB PBOs, respectively. For long term evolution (LTE) 20 MHz 64 quadratic amplitude modulation (QAM) signal, this DPA achieves 29.4 dBm average output power, 18.5% average PAE at $-$ 25 dB error vector magnitude (EVM) limit, and it realizes almost 60 dB dynamic power range from $-$ 30 to 29.4 dBm with EVM below $-$ 25 dB at 2.9 GHz.

In this letter, we report a novel biomemristor synapse based on renewable guar gum biopolymers that can dissolve in water easily and obtain a uniform and compact resistive switching layer via spin-coating process. Incorporating with well-dispersed tungsten telluride (WTe2) nanosheets exfoliated by natural tannic acid, the fabricated Ag/guar gum-WTe2 nanohybrid/Au biomemristor exhibits a forming-free volatile switching characteristic with low operation voltage (< 0.3 V) and robust uniformity. The device can mimic bio-synaptic functionalities for both short- and long-term plasticity, including excitatory postsynaptic current, paired-pulse facilitation, the transition from short-term plasticity to long-term potentiation, and the forgetting behavior modulation by pulse frequency. The guar gum-WTe2 nanohybrid-based biomemristor paves a promising way to develop biocompatible and environment sustainable artificial synapses for neuromorphic systems.

NbO $_{\textit{x}}$ materials have shown considerable potential for the applications of artificial neurons due to their volatile threshold switching (TS) behavior, fast switching speed, and low power consumption. The TS can be attributed to its negative differential resistance (NDR) effect associated with heat accumulation. Experimental data show that the degradation of a NbO $_{\textit{x}}$ -based neuron device is manifested by the shift in transition voltages, including threshold and hold voltages, and shrinkage in the voltage window. This problem strictly undermines the device reliability, and meanwhile, the mechanism is yet to be confirmed. In this work, a physical model involving the role of oxygen vacancies (Vos) is developed to further elucidate the TS behavior in NbO $_{\textit{x}}$ -based devices. For the first time, a continuous drift of the operation voltages upon unipolar biases is simulated and attributed to the redistribution of Vos. We propose and experimentally demonstrate an efficient method that can mitigate the Vo migration problem, leading to a great improvement in the device endurance. Our developed model is not only closer to the real nature of the TS process in NbO $_{\textit{x}}$ but can also provide more insight into the performance optimization of this important class of devices.

NbO $_{\textit{x}}$ -based devices exhibit intriguing promise for beyond-CMOS applications due to their dynamic threshold switching (TS) and negative differential resistance (NDR) behaviors. However, an in-depth study on the degradation scheme of such a device is absent. In this work, we investigate the degradation behavior, i.e., the shift of switching voltages ( $\textit{V}_{\text{th}}$ , $\textit{V}_{\text{hold}}$ ) and the shrink of voltage window (VW), of a nanoscale forming-free TiN/NbO $_{\textit{x}}$ /TiN memristor. Through electrical tests and random telegraph noise (RTN)-based defect tracking, we proved that the shrink of the VW and the increase of switching voltages originate from the increase of electrode resistance due to the oxygen vacancy accumulation. According to the elucidated degradation mechanisms, we propose a reverse refresh strategy to extend the endurance and delay VW degradation. This work provides a possible view of NbO $_{\textit{x}}$ devices’ degradation and may promote the applications.

Vacuum-driven soft grippers are drawing increasing attention in robotics due to their flexibility and adaptability, similar to conventional soft grippers driven by positive pressure. Although better durability and failure safety have been demonstrated, current designs still suffer from high dead weight and limited load-carrying capabilities. In this letter, we present a vacuum-driven gripper consisting of unstretchable fabric chambers and bioinspired elastic spines, capable of compliant, rapid, and powerful grasping. The proposed gripper performs the characteristics of lightweight, high repeatability, and good fatigue resistance. It can exert a maximum grasping force of over 50 N and securely grasp objects of various sizes and shapes, including fast-moving objects. A kinematic model and a quasi-static model are further developed to precisely control the bending angle of the finger, enabling free switching between compliant grasping and squeezing of objects for various application requirements. The proposed design method enriches the field of soft grippers with a simple and replicable approach for achieving safe but high-performance grasping.

The measurements of high-frequency dielectric displacement-electric field hysteresis loops show the continuous reduction of the apparent coercive field upon the lateral size shrinkage of TiN/Hf $_{\text{0.5}}$ Zr0.5O $_{\text{2}}$ /TiN thin-film capacitors. Subsequent measurements of capacitor discharging current transients show the presence of high resistance of an interfacial layer between Hf $_{\text{0.5}}$ Zr $_{\text{0.5}}$ O $_{\text{2}}$ (HZO) and TiN that increases the apparent coercive field significantly for a fast-operating memory. However, the resistance reduces almost linearly upon the size shrinkage, enabling the apparent coercive field reduction from 4.7 to 1.6 mV/cm for the capacitors in sizes ranging from 500 to 14 nm at an operation time of 1 ns. The reduced coercive field can lower operation voltages and improve the cycling number of the memory significantly before the occurrence of dielectric breakdown.

In this study, interface and oxide traps in relation to body-biased hot carrier degradation (HCD) in 14 nm nFinFETs are investigated. An accelerated degradation of device performance is observed as a function of body bias voltage. In particular, the interface traps show a rapid response even at low body bias levels, while the oxide traps show an increase only above a certain body bias threshold. The location of the newly generated traps is meticulously analyzed and the quantized oxide traps are characterized using pulsed I–V measurements. In addition, a comprehensive analysis of the corresponding carrier transportation mechanism is built based on the trap location.

This article presents the design and implementation of a compact CMOS RC frequency reference. It consists of a frequency-locked loop (FLL) that locks the period of a voltage-controlled oscillator (VCO) to the time an RC network takes to charge to a reference voltage. Conventionally, an RC time constant with a near-zero temperature coefficient (TC) is realized by using a trimmed network of resistors with different TCs. In this work, such a network is used to realize a temperature-dependent reference voltage whose TC cancels that of a single-resistor RC time constant. Compared with the conventional approach, which requires resistors with TCs of opposite polarity, the proposed approach can be implemented with resistors with TCs of similar polarity, and so it can be implemented in most CMOS processes. To compensate for RC spread, a trimmed capacitor is used to adjust the nominal frequency. Two prototype chips were made, one based on p-/n-polysilicon resistors and other based on silicided/p-diffusion resistors. Fabricated in a standard 180-nm CMOS technology, the polysilicon-based prototype has an active area of 0.01 mm $^{2}$ and an absolute inaccuracy of $\pm$ 2800 ppm from $-$ 45 $^{\circ}$ C to 125 $^{\circ}$ C with a fixed TC-trim and a one-point frequency trim. After one week of accelerated aging at 150 $^{\circ}$ C, however, significant drift (5000 ppm) was observed. The diffusion-based prototype exhibits greater inaccuracy ( $\pm$ 14 400 ppm) but much less drift (600 ppm).

Esophageal carcinoma (EC) is a common malignant tumor of the upper digestive tract worldwide. An analysis of the latest data from cancer centers in China showed that the incidence of EC and the number of deaths due to EC in China in 2015 were 266,000 and 188,000, respectively, ranking sixth (6.3%) and fourth (8.0%) among all malignant tumors. The early diagnosis and treatment of EC and standardized diagnosis and treatment are important tasks for EC healthcare professionals in various centers across the country. At present, the 8th edition of the EC staging system jointly released by Union for International Cancer Control (UICC) and American Joint Committee on Cancer (AJCC) is the most recent, authoritative and widely used EC staging standard. The EC professional committee of the Chinese Anti-Cancer Association also organizes the "EC Standardization Campaign in China" every year to promote the development of EC diagnostic and treatment norms throughout the country. Since 2011, the EC Committee of the Chinese Anti-Cancer Association has published the Guidelines for Standardized Diagnosis and Treatment of EC. Considering the increasing number of EC clinical studies and the continuous progress in diagnostic and treatment technologies in recent years, the updated Guidelines will include the latest progress in the diagnosis and treatment of EC, with a goal of promoting the forward development of EC diagnosis and treatment in clinical practice.