Search

Gene sequencing technology is a tool which greatly impacts modern biology and medicine. The next-generation sequencing (NGS) lies at the heart of gene sequencing for its massively increasing throughput, but it is difficult to analyze the large quantities of fluorescent images with high accuracy because the fluorescent signals are weak with varying noise signals, and current designs are limited on accuracy and speed. In this paper, we proposed a novel deep learning based gene sequencing pipeline with semi-automatic labelling method. The obtained results are promising, especially on the high-density data, as the BaseFormer surpasses the traditional methods in terms of cluster quality (Q30: 88 %), throughput (16.5% better), and with similar and low error rate (down to 0.137% on average, best at 0.068 % on high-density data).

Due to the need for larger bandwidth, future mobile networks may primarily operate over millimeter-wave (mmWave) bands. In this regard, one of the central research topics today is mmWave channel modeling. However, highly mobile vehicle-to-vehicle (V2V) mmWave channels pose an open question because conducting measurements may be challenging in this environment. The existing publications on mmWave V2V channel modeling consider line-of-sight (LoS) or single-blocker scenarios. Accordingly, the impact of interference from adjacent vehicles is not taken into account. Conventional analytical models suggest that path loss (PL) strikes equally in all directions whereas measurements are inherently directional, especially in the case of mmWave propagation. Recently, it was proposed to synthesize the omnidirectional PL from directional measurements. However, with this method, the resulting PL is underestimated. Moreover, theoretical models assume that vehicular transceivers follow a certain predetermined distribution, which may not be accurate in real-world scenarios. Several works take into account the directionality of an antenna pattern when the antenna orientation is known. However, the effects of reflection, diffraction, and transmission through obstacles are not considered. One way to avoid these limitations is to use ray-based simulations. However, the ray-launching (RL) approach does not specify any criteria for the directionality of the antenna patterns. Assuming omnidirectional PL appears impractical for mmWave transmission where directional communication is crucial to combat high PL. In this paper, we propose an approach for synthesizing directional PL based on extensive RL modeling. For the obtained directional and omnidirectional PL, we parametrize the close-in (CI) and float-intercept (FI) channel models. As a result of our modeling, from 0.1 to 2.7 for the LoS and from 0.1 to 2.6 for the non-LoS (NLoS) higher PL exponent is observed as compared to the omnidirectional case. publishedVersion

In the analysis of ECG data, it is necessary to first automatically classify the quality of ECG data to facilitate the subsequent processing. PhysioNet/Computing in Cardiology Challenge 2011 and existing researches involve the quality of the entire record of ECG, which fails to maximize utilization of ECG data. In fact, single-lead ECG has been widely used in many fields, such as wearable devices, sleep apnea monitoring, and deriving respiratory rate. However, the present methods for evaluating the data quality by each channel only divide them into two categories: acceptable and unacceptable, which is relatively coarse. This paper proposes a new method for the automatic classification of ECG data quality by channel. This method divides them into four categories: (1) electrode shedding, marked as C3; (2) serious noise interference, under which it is difficult to detect R wave, marked as C2; (3) partial noise interference, under which part of R waves may not be detected correctly, marked as C1; (4) high quality signal, marked as C0. The 2011 competition data was re-marked according to the channel with the help of our designed auxiliary program. This paper defined some features and designed a tree classifier using One-Class Support Vector Machine(OCSVM). The test results of our method show that the detection accuracy of electrode shedding is 93.22%, serious noise interference is 90%, partial noise interference is 89.22%, and high quality signal is 97.19%. It shows that the method has a broad prospect in the automatic preprocessing of ECG data.

Classic computer vision algorithms, instance segmentation, and semantic segmentation can not provide a holistic understanding of the surroundings for the visually impaired. In this paper, we utilize panoptic segmentation to assist the navigation of visually impaired people by offering both things and stuff awareness in the proximity of the visually impaired efficiently. To this end, we propose an efficient Attention module -- Lintention which can model long-range interactions in linear time using linear space. Based on Lintention, we then devise a novel panoptic segmentation model which we term Panoptic Lintention Net. Experiments on the COCO dataset indicate that the Panoptic Lintention Net raises the Panoptic Quality (PQ) from 39.39 to 41.42 with 4.6\% performance gain while only requiring 10\% fewer GFLOPs and 25\% fewer parameters in the semantic branch. Furthermore, a real-world test via our designed compact wearable panoptic segmentation system, indicates that our system based on the Panoptic Lintention Net accomplishes a relatively stable and exceptionally remarkable panoptic segmentation in real-world scenes., Comment: 6 pages, 4 figures, 2 tables

In the computational process of the convolutional neural network (CNN) and high-performance computing, the convolution operation dominates the hardware costs and performance of the overall system. To improve the hardware efficiency, the dot-product units (DPU) have been used for convolution computation. Besides, multiple-precision floating-point (FP) support is essential for the accuracy requirement of various applications. In this work, multiple-precision FP many-term DPU is designed with single instruction multiple data (SIMD) structure. The proposed design also supports multiple-precision operations with configurable multiplier and alignment shifter. FP16 twenty-term, FP32 five-term or FP64 one-term dot product operations can be executed in two successive clock cycles without idle multiplication resources. To speed up the summation process, a carry-select adder (CSLA) is designed with excellent area-delay product (ADP) and power-delay product (PDP) performance. The proposed design is realized in UMC 55-nm process with experiment results. Compared with the state-of-the-art multiple-precision work, the proposed design achieves maximum 3.76 times power performance improvement for FP16 operations. Compared with previous CSLA designs, the proposed work improves ADP and PDP performance by 4.7% and 3.91%, respectively.

The high propagation losses and sensitivity to link blockage naturally require dense deployments of millimeter-wave (mmWave) 5G New Radio (NR) systems. One of the inherent challenges in these deployments is cost-efficient backhauling. Addressing this issue, 3GPP has recently proposed the concept of integrated access and backhaul (IAB) to reduce the deployment costs by enabling wireless backhaul. The efficient utilization of spectrum in these systems is conditional on the ability of IAB nodes to simultaneously receive signals on their sectoral antennas. In this paper, we investigate the interference caused by this functionality and identify countermeasures including angular and spatial diversities. Our numerical results demonstrate that the angular distance of 25° between the user equipment (UE) served by adjacent sectoral antennas is sufficient to efficiently mitigate interference. A comparable reduction in the interference level can also be achieved by utilizing spatial diversity with antenna separation of at least 20 m. By combining these methods, one can identify the target levels of angular and spatial diversities suitable for the particular deployment restrictions. acceptedVersion

Near-zero index (NZI) media has the characteristics of local field enhancement, local tunneling and tunneling effect. Here a humidity sensor based on near-zero permittivity (ENZ) media is proposed, which uses the substrate integrated waveguide (SIW) technology. The tunnel height of the designed sensor is much smaller than the height of the SIW structure. It can be observed that there exist strong electric fields in the ENZ tunnel. Placing humidity-sensitive materials in the structure can realize a high-sensitivity humidity sensor. When the humidity changes, it is observed that the frequency of the sensor has changed, and its sensitivity is 1.07 MHz/RH, which proves that the sensor can effectively measure the changes in environmental humidity.

Packet dispatching forwards arrival traffic from the network interface card (NIC) to the virtual network functions (VNFs) in each server in network function vir-tualization (NFV) platforms and has been recognized as a fundamental problem. Traditionally, the dispatching has been CPU-intensive, while existing offloading approaches rely on expensive software-defined network (SDN) switches or Smart-NICs. Our paper proposes FDispatcher, an NFV framework that leverages the Flow Director (FDir) feature on the inexpensive off-the-shelf NICs, to bind flows with destination VNFs for CPU-bypassed packet dispatching. However, the number of bindings that FDir supports is far less than the flow concurrency that an NFV middlebox handles. We divide packet dispatching into FDir-assisted fast path and software-based slow path, and then enable fast-path flows monitoring, reduce FDir reconfiguration overhead, and eliminate potential heavy I/O during scaling. FDispatcher improves packet dispatching throughput by 1.6x~2.8x with a 38x driver efficiency and costs no more than 2 ms for load balancing during scaling.

There is an emerging need to optimize floating-point (FP) dot product units (DPU) for high-performance scientific computing as well as training deep learning models. Due to different precision requirements of applications, a reconfigurable multiple-precision DPU operation can largely reduce the cost of area and power. However, the existing methods could result in redundant bits for unit multipliers, but also leave idle hardware resources for the operations in different precisions. In this paper, a reconfigurable multiple-precision FP DPU design is proposed for high-performance computing (HPC) applications. The FP DPU can be reconfigured as follows. A bit-partitioning method is provided to minimize the redundant bits with a configurable mixed-precision multiplier for three-mode operations: 20 half-precision Dot Product (DP), 5 single-precision DP, and 1 double-precision DP operations. Any of the modes can be executed in two successive clock cycles without idle hardware resources. The proposed design is realized by using the UMC 55-nm process with simulation results. Compared with the existing multiple-precision FP methods, the proposed DPU achieves 88.9% and 35.8% area-saving performance for FP16 and FP32 operations, respectively. Moreover, when using benchmarked HPC applications where multiple precisions can be used, the proposed reconfigurable DPU can accelerate up to 4× and 20× maximum throughput rates when compared with fixed FP32 and FP64 operations, respectively.

Tripping or falling is among the top threats in elderly healthcare, and the development of automatic fall detection systems are of considerable importance. With the fast development of the Internet of Things (IoT), camera vision-based solutions have drawn much attention in recent years. The traditional fall video analysis on the cloud has significant communication overhead. This work introduces a fast and lightweight video fall detection network based on a spatio-temporal joint-point model to overcome these hurdles. Instead of detecting falling motion by the traditional Convolutional Neural Networks (CNNs), we propose a Long Short-Term Memory (LSTM) model based on time-series joint-point features, extracted from a pose extractor and then filtered from a geometric joint-point filter. Experiments are conducted to verify the proposed framework, which shows a high sensitivity of 98.46% on Multiple Cameras Fall Dataset and 100% on UR Fall Dataset. Furthermore, our model can achieve pose estimation tasks simultaneously, attaining 73.3 mAP in the COCO keypoint challenge dataset, which outperforms the OpenPose work by 8%.

Temperature is the key parameter of the operating state in the power system. The occurrence of faults of high-voltage transformer, cable joint and other power equipment is often accompanied by abnormal temperature rise. Wireless monitoring of the temperature can effectively prevent accidents and ensure the safe and stable operation of the equipment. Here two kinds of passive wireless temperature sensors based on the effective localized surface plasmons or near-zero-index (NZI) medium have been demonstrated. The simulation results show that the passive wireless sensors can effectively perceive the temperature change and verify the feasibility of the sensors.

To solve the problem of waviness on the end surface of the transmission housing, the LabVIEW platform was adopted with the help of acceleration sensors and PCI (peripheral component interconnect) acquisition card. The vibration frequency range and the vibration distribution of the transmission housing in excitation response was determined using finite element modal analysis and harmonic response analysis. Then the layout of the transmission housing stiffener was optimized combining actual processing diagnostic test. In this way, the vibration characteristics of the transmission housing could be improved in the machining process, the vibration frequency could be offset to the natural frequency, and the vibration of the machine could be reduced effectively. The test results show that the machining vibration of the transmission housing could be greatly reduced in the new stiffener layout, and the surface quality and the machining performance of the transmission housing could be ensured. Therefore, the proposed method is practical for the structural optimization of the parts that are prone to vibration. By combining experimental test with finite element simulation, the new approach provides a reference to the future study on vibration characteristics, vibration fault diagnosis, and dynamic characteristics of milling structure system.

In this work, we have explored the automatic design of an operational amplifier using multi-objective optimization technique. The design parameters of the circuit, including transistor sizes, bias current and load capacitance, have been optimized by genetic algorithm, in order to achieve the best compromise (Pareto-Front) between performances, such as Gain, cut-off frequency, quiescent current and phase. Optimizations with two and three objectives have been demonstrated successfully. Various population size and optimization algorithms have been used in our work and compared systematically.

Due to the deformable and locally adjustable characteristics and local field enhancements, near-zero-index (NZI) medium has found applications in sensors, waveguides, integrated circuits, and other fields. Here a cylindrical microfluidic channel is etched into epsilon-near-zero (ENZ) medium and a microfluidic sensor with high sensitivity is realized The transmission coefficients and phase of the magnetic fields are observed. It can be seen that the sensor has high sensitivity and small changes of the effective permeability can be effectively measured.

5G new radio (NR) requires a very dense deployment of cellular infrastructure, which poses a great economical challenge if traditional fiber backhaul links are used. To cope with this problem, 3GPP introduced integrated access and backhaul (IAB), aiming to use wireless backhaul to provide high capacity and great deployment flexibility. The multi-hop and multi-route topology for this new type of access networks challenges the existing retransmission-/repetition-based reliability-enhancing technique, such as ARQ/HARQ and packet duplication. However, it also opens up new opportunities for novel reliability enhancement techniques. In this paper, taking advantage of the more complex network topology of IAB, we propose to use linear network coding as a potentially better solution to improve end-to-end latency and reliability. We discuss its placement in the IAB protocol stack, and propose two novel schemes to improve the performance of network coding in the IAB network: the rate-proportional traffic splitting scheme in the multi-route scenario, and the adaptive coded-forwarding scheme in the multi-hop scenario. Simulation results show that in some typical IAB scenarios our network coding solution has considerable performance gains over the existing repetition-based technology.

In recent years, with the development of ultra-high speed aircraft, rocket, missile and artificial earth satellite technology, higher requirement have been put forward for radar’s coverage of airspace, target tracking ability and multi-target tracking technology. Phased array radar has gradually replaced the traditional mechanical scanning radar due to its excellent tracking performance, which plays an important role in more specific fields. In this paper, the influence of different factors on the gain pattern of phased array antenna is analyzed in detail using theoretical calculation and simulation research methods for several types of phased array antenna. Research shows that the factors such as the number of array elements, the spacing of array elements and the scanning angular position determine the maximum value, main lobe, grating lobe, side lobe and the zero point of the gain pattern. Finally, a parameter calculation scheme is designed, which has certain reference value for the design of phased array antenna.

As of today, mmimeter-wave (mmWave) bands are employed as part of the emerging 5G technology to provide high data rates for vehicle-to-vehicle (V2V) communications. However, V2V channels over mmWave have not been well studied as of yet due to the complexity of measurements, especially if there is a need to estimate the contribution of adjacent interfering vehicular transceivers. Moreover, the channel models that are currently in use consider vehicles on the road according to a certain distribution, which may not be accurate in practice. Also, past models do not take into account the effects of reflection, diffraction, and transmission through obstacles, as well as the physical properties of vehicles themselves. In this paper, using geometric ray-based simulations, in which the aforementioned effects are incorporated, we present mmWave V2V channel modeling for a highway scenario at 28 and 72 GHz carrier frequencies with both low and high density of vehicles. Our results include such characteristics as path loss, fading, root-mean-square (RMS) delay spread, and angular spread. acceptedVersion

Different from traditional single-element and phased-array antennas, frequency diverse array (FDA) uses a slight frequency offset across the elements to produce angle-, range- and even time-dependent transmit beampattern, which provides promising potentials to develop new jamming techniques on synthetic aperture radar (SAR) imaging. In this paper, we propose a novel deception jamming method against SAR imaging, which uses FDA to quickly generate false targets. Moreover, the position of the false target is controlled by the jammer, and the number of false targets is determined by the number of FDA elements. The proposed methods are validated by extensive simulation results with range-Doppler imaging algorithm.

The weighted structure tensor (ST) can be usually used on edge detection of multi-channel images, including polarimetric synthetic aperture radar (polSAR) images. The traditional weighting method is average weighting, each channel considered to provide the same edge information. But in fact different channels tend to contain different amount of edge information, so the traditional weighting method cannot be adopted to extract full edge information of the multi-channel images. This letter proposes a novel weighting method for ST, in which the weight of each channel is obtained by the eigenvalue-measured way. Experimental results show that the proposed method outperforms the traditional method on edge detection.

This paper proposes a simple two-dimensional imaging via stationary nonuniform frequency diverse array (FDA) transmitter by exploiting FDA both range and angle dependent transmit beampattern. The nonuniform FDA transmitter only generates range-dependent beampattern, so that two-dimensional (range-angle dimension) imaging of targets can be achieved without the employment of moving radar platform, by jointly utilizing the only range-dependent transmitter beampattern and general single-antenna or phased-array antenna. The effectiveness of the proposed method is verified by simulation results.

Nowadays, IoT video applications are in a sharp rise, various real-time video streaming of video surveillance systems transmitted via Internet are widely investigated. The real-time video surveillance can actively monitor and detect the abnormal events in time. In 5G HetNets, we specifically develop a compressed sensing based linear predictor to predict the traffic load at the next moment. The results justify that our proposed method can forecast the traffic load and improve system performance.

China's “pain of lack of core” objective facts, the party central committee is very concerned. Especially after the US sanctions on ZTE this year, everyone in China is thinking about the “China core”: the Internet power cannot be without the “China core”, the network power more indispensable “China core”, digital China more need “China core”! How does“China core” re-encircle breakthrough, how to build? It has become a major proposition that our government, IT community, and many scholars pay great attention to. This paper uses external theory as guidance, draws on the successful experience of foreign integrated circuit technology alliance, and discusses the construction factors of China's general chip industry technology alliance, in order to provide reference for solving the independent supply of “China core”.

Recently, the user-side data of power grids gradually exhibit massive data and high complexity features. The traditional power anomaly detection model has been difficult to meet the existing requirements. In recent years, the neural networks and machine learning methods that are widely used in anomaly detection, but all of these methods have a high demand for training samples and cannot be well applied when missing sample tags on the power data set. This paper designs an unsupervised power data anomaly detection model mainly based on the isolated forest algorithm. The model includes modules for feature extraction, feature reduction, and isolated forest computing. The research results show that using this model to detect abnormal power usage data can process large amounts of data in a short time, but also can accommodate the lack of training samples and can better meet the practical needs of the power sector.

In this paper, a novel planar mode-transducing antenna is proposed. The proposed antenna could accept directly microwave energy in an azimuthally symmetric mode and radiate a high-gain linear polarized mode, which is a good potential candidate for the high-power microwave system.

For speech emotion recognition, emotional feature set with high dimension may produce redundant features and influence the recognition accuracy. To solve this problem and obtain the optimal emotional feature subset of speech, a feature dimension reduction based on linear discriminant analysis is proposed. According to the confusion degree between different basic emotions, an emotion recognition method based on support vector machine decision tree is proposed. Experiment on speaker-dependent speech emotion recognition using Chinese speech database from institute of automation of Chinese academy of sciences is performed and a speech emotion recognition system is presented, where standard feature sets of the INTER-SPEECH and classic classifiers are used in comparative experiments respectively. Experimental results show that the proposal achieves 84.39% recognition accuracy on average. By proposal, it would be fast and efficient to discriminate emotional states of diverse speakers from speech, and it would make it possible to realize the interaction between speaker and computer/robot in the future.

Test flight of a new commercial aeroplane is crucial in validating the functionality, safety and performance of the new aeroplane design before its batch manufacturing can take place. Massive amounts of data streams are typically generated from thousands of sensors on an aeroplane during test flight, which require realtime processing to detect anomaly and to predict malfunctions for emergency response. This paper provides an overview of recent research in custom machine learning architectures which have shown promise for highspeed data processing, and proposes a time series learning model based on LSTM (Long Short Term Memory). This LSTM model is adopted for realtime data analysis used in anomaly detection for the COMAC C919 test flight. A custom architecture targeting FPGA (Field Programmable Gate Array) implementation for the proposed approach can be embedded into realtime data analysis and processing platforms for large commercial aircraft.

In the field research area, a large-scale automatic data collection is always in a great need. However, traditional communication methods used to support the data collecting activity is less efficient due to the complex deployment and limited ability to handle large amount of devices. Furthermore, the deployment of observation station highly depends on telecommunication carriers where there is carrier's network coverage. In terms of the utilization of the collected data, there is a lack of a platform that enables data sharing among various research projects and areas, which may significantly limit the value of data. Therefore, this paper proposes a data collection infrastructure specifically designed for observation stations in field research activities based on Low Power Wide Area Network technology and Internet of Things name service platform. This infrastructure could help to build large-scale data collection networks while avoiding telecommunication carriers' restriction, as well as provide a deployable solution to fully exploit the value of data.