To overcome the challenges posed by domain shift in hyperspectral image (HSI) classification, methods based on domain adaptation (DA) have been widely used. Currently, most HSI DA methods focus on designing complex strategies to align the distributions of the source domain (SD) and the target domain (TD) in the feature space after feature extraction, yielding promising results. However, when there exists a large domain shift between SD and TD, it becomes challenging to map them into the same feature space. In this article, we propose the bidirectional mamba and domain mixing network (BMDMnet). Since pure CNN architectures are constrained in local feature extraction, while transformer-based models improve global feature capturing capability at the cost of high computational complexity, we propose the bidirectional mamba module (BMM) as an efficient solution for capturing long-range dependencies. In addition, a self-distillation strategy is employed during training. By utilizing a more stable teacher model, reliable predictions can be obtained in the TD. Subsequently, a domain mixing supervised learning (DMSL) module is designed, which creates a mixed domain by selecting low-entropy sample-pseudo-label pairs from the TD and randomly combining them with sample-label pairs from the SD. DMSL aims to introduce mixed domain to mitigate the inter-domain gap in the data space, thereby enabling the model to learn TD representations more effectively. Experiments demonstrate that BMDMnet outperforms state-of-the-art algorithms across three cross-scene datasets.

Utilizing hyperspectral remote sensing technology enables the extraction of fine-grained land cover classes. Typically, satellite or airborne images used for training and testing are acquired from different regions or times, where the same class has significant spectral shifts in different scenes. In this paper, we propose a Bi-directional Domain Adaptation (BiDA) framework for cross-domain hyperspectral image (HSI) classification, which focuses on extracting both domain-invariant features and domain-specific information in the independent adaptive space, thereby enhancing the adaptability and separability to the target scene. In the proposed BiDA, a triple-branch transformer architecture (the source branch, target branch, and coupled branch) with semantic tokenizer is designed as the backbone. Specifically, the source branch and target branch independently learn the adaptive space of source and target domains, a Coupled Multi-head Cross-attention (CMCA) mechanism is developed in coupled branch for feature interaction and inter-domain correlation mining. Furthermore, a bi-directional distillation loss is designed to guide adaptive space learning using inter-domain correlation. Finally, we propose an Adaptive Reinforcement Strategy (ARS) to encourage the model to focus on specific generalized feature extraction within both source and target scenes in noise condition. Experimental results on cross-temporal/scene airborne and satellite datasets demonstrate that the proposed BiDA performs significantly better than some state-of-the-art domain adaptation approaches. In the cross-temporal tree species classification task, the proposed BiDA is more than 3%∼5% higher than the most advanced method. The codes will be available from the website: https://github.com/YuxiangZhang-BIT/IEEE TCSVT BiDA.

Pine wilt disease (PWD) is one of the most prevalent pine trees diseases, resulting in both ecological and economic havoc. UAV remote sensing segmentation plays a crucial role in early identifying and preventing PWD. However, deep learning segmentation models customized for PWD identification in scenarios with complex backgrounds have not received extensive exploration. In this paper, we propose a novel UAV remote sensing segmentation model called Clusterformer with a conventional encoder-decoder structure. The encoder is comprised of the specially designed Cluster Transformer, which includes a cluster token mixer and a spatial-channel feed-forward network (SC-FFN). The cluster token mixer utilizes constructed clusters from the feature maps to represent pixels, thereby reducing redundant and interfering information. The SC-FFN extracts multi-scale spatial information through depth-wise convolutions and channel information through a multilayer perceptron in sequence. The decoder primarily consists of the specially designed D-Cluster Transformer. The token mixer of the D-Cluster Transformer employs constructed clusters from high-level decoded tokens to represent low-level encoded tokens without relying on traditional upsampling methods such as interpolation, transpose convolution, or patch expansion. Consequently, more robust and less redundant features from high-level decoded feature maps are transferred to low-level encoded feature maps. Experimental results on two PWD datasets demonstrate that Clusterformer outperforms existing state-of-the-art segmentation models. This confirms the effectiveness and efficiency of Clusterformer in PWD identification. Code is available at https://github.com/huanliu233/Clusterformer.

Classification of tree species is of great significance to forest surveys. Recently, considering the low differences of spectral information among tree species, enhancing the dependence between long-distance bands has become a research hotspot. A tree species classification method based on a convolutional (2-dimension) long short-term memory (Conv2DLSTM) network and transformer is proposed. First, the main features of HSI are retained by principal component analysis (PCA). Then, the Conv2DLSTM network obtains the global correlation information between long-distance band pixels, and the 3-dimensional convolutional neural network (3DCNN) updates the local spatial-spectral information. Finally, low-level features are converted into semantic tags to guide the modeling of high-level semantic features. The experimental results on the forest dataset demonstrate that the proposed method is superior to other competitive work.

This article introduces the Chinese Terrestrial Ecosystem Carbon Inventory Satellite(TECIS), the first mission with the integration of active and passive sensors for forest carbon mapping. TECIS utilizes time-synchronized multiple-beam LiDAR and multi-angle optical imagery for forest carbon monitoring. We first provide an overview of the satellite's features and discuss the observational capabilities of the LiDAR and multi-angle payload. The preliminary results for forest height estimation analysis were shown using the payloads. The Bidirectional Reflectance Distribution Function (BRDF) features such as hot/dark spot information, were calculated based on the multi-angle images. A deep learning approach for forest parameter estimation through the fusion of LiDAR and BRDF data.

A Q-learning based current-steering digital to analog converter (DAC) calibration method is proposed in this paper for spurious-free dynamic range (SFDR) improvement. A look-up table (LUT) to control the switching sequence of the DAC elements is achieved by Q-learning for an optimal SFDR. Compared with the fixed element transition strategy proposed by manual derivation, the LUT can be updated by off-line training to deal with diverse and complex non-ideal factors limiting the SFDR of DAC. In this paper, a 2.0-GS/s 12-bit segmented current-steering DAC in 28nm process is simulated and the equivalent model is extracted to verify the effectiveness of the method. Simulation results show that, the SFDR over entire Nyquist bandwidth is larger than 70 dB with about 8 dB improvement using the proposed Q-learning method.

A resource-efficient neural-network-based face detector using 1.5-bit frame-to-frame delta quantization with diagonal spatial feature extraction method is proposed in this paper, which is designed for resource-limited always-on camera sensors. The proposed architecture completes analog-domain frame difference for motion sensing, which triggers digital-domain feature extraction. Based on the sparse and effective features, a lightweight convolutional neural network is devised as a classifier. A self-recorded dataset of 313 videos for humans of different appearances, light intensity and backgrounds is used to validate the performance of the proposed method. Simulation results show that the proposed method achieves 93.6% accuracy using only a$\boldsymbol{50\times 50}$pixel array, which is higher than the prior discontinuous temporal change quantization method. Meanwhile, the conservatively estimated power consumption of the proposed method can be reduced by$\mathbf{14 \times}$compared to the state-of-the-art work.

A Ternary neural network (TNN) based patient- specific single lead Electrocardiography (ECG) processor for the early detection of cardiac arrhythmias (CA) is presented. The designed system detects upward/downward turning points in the ECG to detect the slope variation and calculates the fiducial points of the PQRST beats, with high auto-patient adaptability. A 3-layer Feedforward Neural Network with ternary weights is integrated on the sensor to classify eight different types of Shockable CA (SCA) and non-SCA (NSCA) with sensitivity and specificity of 99.1% and 99.8% respectively. The proposed processor is also synthesized using 65nm CMOS technology having an area of 1.08 mm2with an overall power consumption of 2.52 μA, energy efficiency of 72 nJ/detection.

Motivated by the ultra-low power design demands of the integrated implantable medical electronics and systems, a switched-capacitor DC-DC converter is presented in this paper. The DC-DC converter is controlled by hysteretic control approach, which has inherently stable and fast loop speed compared with the approach using error amplifier. To achieve higher power efficiency and lower quiscent current, in addition to optimizing the switching frequency and width of the core module of the converter, Low power design methodes are used for the functional module. By operating transistors in the subthreshold region and decreasing operating frequency, the short current and dynamic power can reduce a lot. The converter has designed in 0.18μm standard CMOS technology. It can achieve step down voltage conversion of 1.2V with a ripple voltage of 1.4mV from 2.5V input voltage. A high power efficiency of 88.3% can be achieved at a load current of 3μA. It also produces about 34nA low quiescent current without load, which is suitable for the integrated implantable biomedical electronics.

Impedance matching plays an important role in radio frequency circuits. To achieve dynamic load modulation in back-off operation for power amplifier or address the variation of antenna impedance due to complicated environment, tunable matching networks are demanded. To optimally design the coverage area within the Smith Chart for common ladder networks, analytic calculating procedure and computer optimization method has been proposed. The calculating procedure can be applied to arbitrary ladder networks and the optimization algorithm can acquire the optimal network automatically with appropriate constraints.

This paper presents a fully-integrated 8 MHz relaxation oscillator for ultra-low-power applications. Fabricated in 0.13 μm CMOS process, the oscillator occupies an area of 0.022 mm2. With adaptive reference generation and low-swing oscillation design, the power consumption is 2.8 μW, resulting a figure-of-merit of 0.35 nW/kHz. The adaptive reference feedback compensates the comparator delay and filters out the low-offset frequency part of the noise and temperature variation. To reduce the frequency influence due to current mismatch on the reference voltage and oscillation node, cascode current mirrors are applied. To reduce the influence due to the delay of capacitor reset logic, a pulse-to-edge generation block is used. The measured temperature stability is 1.24% in the temperature range of -20 °C to 60 °C and the measured periodic rms jitter is 180 ps, 0.144 % jitter-per-period on the output.

Human bowel sounds (BSs) deliver much useful information about gastrointestinal health status. In recent years, the utilization of advanced bio-acoustic sensors, such as the wired electronic stethoscopes and the wireless wearable sound recording patches, has enabled researchers to record, store and analyze the BSs in digitized manners, i.e., computerized bowel sound analysis. However, for collected BS recordings, how to effectively pick up the segments that contain BS events while ignoring those segments that only contain background noises remains difficult. Moreover, the BS segment recognition algorithms that are meant to be applied in the wearable healthcare scenes are further required to retain a low computation complexity. In this work, a light weighted BS recognizer based on convolutional neural networks (CNNs) is proposed for wearable systems. Specifically, the proposed recognizer firstly converts each one-dimensional segment into a two-dimensional spectrogram by calculating the Mel-frequency cepstrum coefficients (MFCCs) frame by frame and then passes the spectrogram through a CNN to infer the category of the segment. To validate the CNN-based BS recognizer, a 28 minutes of BS dataset that contains 955 BS-present segments and 725 BS-absent segments are made. Experimental results on the dataset show that the recognizer attains the 91.25% and 90.83% mean accuracies for the `not-across-subjects' and `across-subjects' validation, respectively. Moreover, compared with the state-of-the-art LSTM approach, the CNN-based BS recognizer is light-weighted with only 20.35k parameters, which is a quarter of that of the LSTM. Due to the lower model complexity, the CNN-based BS recognizer has the potential to be integrated into the gateways in a wearable system to secure the function that only the BS-present segments are allowed to be relayed to the remote. User privacy can be better protected in this way.

A wearable ECG monitoring device with accelerometer aided activity recognition is proposed in this work. A 3-axis accelerometer is integrated in the Band-Aid alike wearable device which will be stuck to the user's chest. The Lead V2 ECG signal and the chest acceleration data are recorded synchronously. An activity recognition algorithm is proposed to identify certain types of daily activities, including coughing, walking, standing, sitting, squatting or lying based on the chest acceleration data. The recognition result can be further used to correlate the recorded ECG signal to the user's activity. Experiments on 13 volunteers with age of 5 to 68 show that the proposed algorithm have an overall accuracy of 96.92%. The recognition result can be further used to correlate the recorded ECG signal to the user's activities.

Hyperspectral remote sensing data have great potential to identify ground objects and classify tree species. However, the tree species classification based on hyperspectral data in the subtropical region of hilly landscape have always been challenged by rugged topography. We conducted our research in subtropical mountainous forests in Pu'er of Yunnan province in southwestern China. This research investigated the capability of airborne AISA Eagle II hyperspectral data in tree species classification, especially in mountainous areas. Integrated Radiometric Correction (IRC) is applied as atmospheric and topographic correction technique; a classification algorithm was developed based on the topographic-corrected data. Principal Component Analysis (PCA) method was used to reduce the dimension of hyperspectral data before classifying tree species. The first three components indices combined with texture features were used for Support Vector Machine (SVM) classification. Study results demonstrated that this developed method obtained a good performance in detecting the target tree species for the overall classification accuracy is 95.14% and kappa coefficient is 0.93.

A campaign aimed at airborne multi-angle hyperspectral data collection was designed and implemented in Daxinganling of Northeastern China in the August of 2015. The purpose of this campaign is to investigate the potentials of multi-angle optical data for forest height and biomass estimation. The observation angles of 0°, 20°, 44° and 55° were used with 400-1000 nm spectrum range. To provide comparable information, airborne Lidar data were collected simultaneously. First results showed the collected angular information captured the Bidirectional Reflectance Distribution Function (BRDF) characteristics of forest spectrum. The relationship of lidar derived vertical parameters and multi-angle optical data are under further investigation.

China has a wide variety of forest types. It is challenging to make a reliable estimation of these forest aboveground biomass (AGB) using geo-spatial technologies. We developed a Field-Airborne-Spaceborne (FAS) comprehensive observation method for AGB estimation. According to forest ecological zones of China, we carried out three FAS campaigns in the Northeast, central, and Southwest of China. Airborne LiDAR data were collected along National Forest Inventory (NFI) plots. Then the airborne LiDAR data were used to estimate AGB after been trained by NFI plots. Then these LiDAR estimated AGB were used to train satellite data for large area biomass mapping. The stratified regression tree modeling method was used in this research. The overall estimation correlation coefficient are better than 0.8.

This paper presents a high precision direct multi-cell Battery Voltage Detecting Circuit (BVDC) for Battery Management Systems (BMS) in electric vehicles. BVDC in BMS must be able to accommodate direct voltage input up to tens of volts from series connected batteries, fulfil the precision needs by battery state of charge estimation algorithm. The BVDC circuit is designed with a 0.5μm 1-poly 3-metal high voltage (HV) 60V Bipolar-CMOS-DMOS (BCD) semiconductor process. System optimization design of HV multiplexer and incremental sigma-delta analog-to-digital converter (ADC) in BVDC circuit eliminates the need of amplifier-resistor based level shift, and any static current through HV signal path, which greatly improved the conversion accuracy. Post layout extracted simulation result shows that the typical conversion error is less than 0.3mV@1MHz for series connected batteries without extra calibration.

Coordinated Multi-Point (CoMP) is an effective way to improve user performance in next-generation wireless cellular networks, such as 3GPP LTE-Advanced(LTE-A). The base station cooperation can reduce interference, and increase the signal to interference and noise ratio (SINR) of cell-edge users and improve the system capacity. However, the base station cooperation also adds additional power consumption for signal processing and sharing information through back-haul links between cooperative base stations. As such, CoMP may potentially consume more energy. This paper studies such energy consumption issue in CoMP, presents a semi-dynamic CoMP cluster division algorithm based on energy efficiency (SCCD-EE) that can effectively adapt to users' real-time interference, and employs the idea of Maximal Independent Set (MIS) to solve the problem of cluster overlapping. To verify the feasibility of the proposed algorithm, this paper performs comprehensive evaluations in terms of energy efficiency and system capacity. The simulation results show that the proposed semi-dynamic cluster division algorithm can not only improve the system capacity and the quality of service (QoS) of cell-edge users, but also achieve higher network energy efficiency compared with static cluster methods and Non-CoMP approaches. Copyright © 2016 John Wiley & Sons, Ltd.

This paper describes an effective way of relaxing the nonlinearity problem of the digitally-controlled oscillator (DCO) in the two-point modulator design. A separate coarse varactor array dedicated for the high-pass modulation significantly simplifies the nonlinearity calibration of the DCO with a few-bit control. In addition, a finite-impulse response (FIR) filter is designed for the multi-bit high-pass modulation path to reduce the quantization noise, while offering a time-interleaving operation to minimize the DCO sensitivity to the coupling during switching time. A two-point modulator based on a semidigital fractional-N phase-locked loop (PLL) is implemented in 0.18μm CMOS. Simulation results show that the proposed modulator can achieve 10Mb/s GFSK/GMSK modulations with the EVM of -37dB.

The fetal heart rate is vital for monitoring fetal well-being. Fetal heart rate monitoring based on acoustic techniques is passive and noninvasive. In this work, a fetal heart rate monitoring system based on phonocardiographic method is proposed. A portable low-power stethoscope is customized which meets the need for sensitivity in the monitoring. A noise cancellation method and adaptive matching method are applied to extract the fetal heart rate effectively. Clinical trials are carried out on pregnant women, and the comparison of fetal heart rates given by the proposed system with those given by the Doppler monitor is given to show the accuracy.

Computing-intensive algorithms which occupy most of executing time are always the main bottleneck in real-time or high quality video applications. In this paper, the optimization methods of the computing-intensive decoding algorithms of H.264, including MC (Motion Compensation), Deblocking and IDCT-IQ (Inverse Discrete Cosine Transform-Inverse Quantization), are proposed firstly, and then implemented on the REMUS (REconfigurable MUltimedia System) which is an embedded coarse-grain reconfigurable multimedia system. Tests show that the efficiency of MC is improved by 32.5%, Deblocking by 69% and IDCT-IQ by 88.5% compared with XPP PACT(a commercial reconfigurable processor). Compared with typical ASIC solutions, execution performance of MC and IDCT improved by 49% and 17%, respectively, while that of Deblocking remained about the same.