Abstract： In order to meet the requirements of future communication such as wide band, large capacity, long distance and multi-member, directional antenna has been gradually and widely used. The directional antenna only sends and receives electromagnetic waves in some specific directions, compared with the omnidirectional antenna, it has good concealment and anti-interference ability, and increases the effective utilization of radiation power, so it can expand the communication distance and adapt to the complex environment of multi-member better. But at the same time, this feature makes “directional antenna alignment” become the hot spot of research. The beam gets narrower in order to alleviate the influence of propagation loss when the communication frequency gets higher, which intensifies the difficulty of alignment. And both communication parties move at a high speed, how to keep the antenna tracking in real time is another crucial. Therefore, aiming at the air-to-air scenario, this paper summarizes the methods of pointing solution, angle measurement and dynamic tracking, and discusses both the advantages and disadvantages of each method. Then the system architecture based on a control scheme combining coordinate tracing strategy and signal tracing strategy aided by motion prediction is concluded, and the potential challenges and future development of directional antenna alignment and tracking technology are also illustrated.

Abstract：Image texture feature extraction is the basis of image texture classification and segmentation, and is widely used in medical images, remote sensing images and other fields. The recent mainstream texture feature extraction methods and fusion methods are analyzed and summarized, and the future research direction of texture feature extraction methods is discussed. Firstly, the characteristics of 12 mainstream public texture datasets and their applicable scenarios are introduced in detail. Then, some texture feature extraction methods that have emerged in recent years are summarized, and the texture feature fusion methods are classified from multiple angles according to the feature fusion methods. Finally, the difficulties and challenges of texture feature extraction are summarized, and the research direction of texture feature extraction methods in the future is considered and discussed.

Abstract：Most traditional deep learning point cloud complement learning methods only use the global features and ignore the local features. In order to better extract and use the local features of point cloud, an end-to-end cloud completion network based on deep learning is proposed in this paper. On the basis of point cloud completion network (PCN), the coding part introduces dynamic graph convolution (DGCNN) improved for local features. The edge convolution of multiple different dimensions is used to extract more abundant local features, and weaken the characteristics of the far point according to the distance. Then the network structure is optimized with the idea of deep residual network connection to achieve the fusion of multi-scale features, and the mean pooling method is added to compensate for the information loss caused by global pooling. In the decoder part, FoldingNet was used to make the output point cloud complete. The experimental results show that the point cloud completion network is partially improved compared with PCN and other point cloud completion networks, which verifies the effectiveness of the new method.

Abstract：In order to solve the problem of high difficulty and large amount of computation in feature extraction for web spam detection, a method for extracting semantic features only based on the HTML script of the current page is proposed. Firstly, the domain name is segmented by a memorization search algorithm combining depth-first search and dynamic programming. Secondly, The latent Dirichlet distribution is used to extract subject words of the web page. Lastly, three single-page semantic similarity features are calculated based on Word2Vec and word mover distance. Combining the single-page semantic similarity features with single-page statistical features, classification algorithms such as random forest are used to build classification models for web spam detection. The experimental results show that the AUC value of single-page content extraction based on semantic and statistical features for classification reaches 88.0%, which is about 4% higher than that of the control method.

Abstract：In order to solve the problem of indiscriminate parking of shared bikes in cities, a standard parking system for shared bikes based on OpenMV was proposed. The system is mainly composed of OpenMV development board, BC20 communication positioning module and OneNET cloud server. Among them, OpenMV completes the detection and recognition of the parking state, and uses the built-in network module of BC20 to conduct data interaction with the OneNET cloud server, and feedback the detection results to the small program client, so as to realize the parking detection and control of shared bikes. Through a lot of tests, the experimental results showed that the system runs stably, and the recognition accuracy rate is 94.1%, which can realize the standard parking automatic detection of shared bikes, and has a good market application prospect and value.

Abstract： In the existing hardware Trojan detection methods, there is problem of low detection rate. Therefore, a cost-sensitive hardware Trojan detection was proposed. By analyzing the structural and signal features of Trojan circuits, an 11 dimensional Trojan feature vector was established. A Trojan feature expansion algorithm based on Borderline-SMOTE was proposed, which effectively expanded the Trojan sample information in the training set. Based on PSO algorithm, the parameters of SVM model were optimized, and a cost-sensitive classification model was established. 17 benchmark circuits from the Trust-Hub were used to verify the efficacy of the proposed approach. Among them, the TPR of 16 benchmark circuits is 100%, and the average TNR is as high as 99.04%. Compared with other existing methods, the detection rate of Trojan is improved greatly.

Abstract： At present, NAND Flash memory is mostly used in the design of high-speed and large-capacity storage device. Aiming at the shortcomings of serial port detection and data reading and writing methods, such as slow detection speed and long waiting time, a NAND Flash detection and control method based on USB2.0 bus was proposed. The communication and control circuit is designed by using FPGA logic control function and high-speed USB interface chip, and the command and operation control of Flash is realized by PC software. The user can complete the reading and writing, erasing detection and bad block calibration of NAND Flash through PC application program. The experimental results show that the method is fast, flexible, accurate and effective in reading, writing, erasing and bad block calibration, and can be widely used in the design and development of storage test devices.

Abstract：To address the problem of system following performance degradation due to load torque variation and the problem of unknown load torque that occurs in DC brushed motors when using conventional fuzzy PID controllers for position following control， a load observer-based improved fuzzy PID controller is proposed to replace the traditional fuzzy PID controller. A load observer is added to the traditional fuzzy control structure in order to observe the system load torque in real time and adjust the parameters of the system controller in real time using the load torque, which improves the position following capability of the system when the load torque changes abruptly. The simulation results show that the load observer can observe the load torque in real time and accurately, and the improved fuzzy PID controller has a strong ability to suppress the load disturbance and can effectively reduce the regulation time and steady-state error of the system.

Abstract：Inductors have many parameters, of which the two most important parameters determine the performance of the inductor: inductance and saturation current. The existing inductance parameter measurement methods can either measure the inductance value more accurately or use other auxiliary means to measure the saturation current, but they cannot measure the inductance value and the saturation current at the same time. Combining the special single trigger and graphic display technology of the digital storage oscilloscope, this paper designs a simple resistance and inductance series measurement circuit. According to the measurement definition of inductance value, it measures the voltage value at both ends of the inductance and the current change value of the sampling resistor to calculate the inductance value. The digital storage oscilloscope displays the inflection point of the measurement curve, and the inductor saturation current can be calculated. It is verified by experiments that this method can obtain accurate inductance values and at the same time obtain intuitive saturation current values. The method is simple in principle, reasonable in design, and easy to operate. It is an economical and effective method for measuring inductance parameters in the absence of an LCR tester in the laboratory.

Abstract： In order to track and locate the moving target in complex indoor environment, a double-layer Kalman filter (DKF) with weakening NLOS noises is designed, which cascades the classical Kalman filter (KF) and the Extended-Kalman filter (EKF). A method for distinguishing the noises is introduced into KF by adjusting the covariance according to the residual between the prediction and measurement. Through this method, the filter gain of KF is able to adjust adaptively, so that the distances measured by ultra-wide band (UWB) sensors can be smoothed and then input into the next EKF. Finally, the real-time positioning is achieved by outputting the position information of the moving target after EKF at each iteration. The algorithm is simulated on MATLAB, and the tracking accuracy is compared with several existing algorithms under the constant velocity (CV) model. The proposed DKF can achieve high accuracy within 3 cm in LOS environment and 10 cm in NLOS environment.

Abstract：Platoon-based driving has been proved to effectively improve road capacity and energy efficiency. On this basis, this paper investigates wireless connectivity of platoon-based vehicular networks with the presence of eavesdroppers. Three types of secure connectivity probability, intra-platoon secure connectivity probability, extro-platoon secure connectivity probability and network secure connectivity probability, are defined and derived to respectively characterize the network constituted by a single platoon, a double platoon and multiple platoons. Simulation results show that system parameters affecting connectivity performance include transmission range, number of eavesdroppers and vehicle density. Particularly, secure connectivity probabilities are maximized by an optimal vehicle density.

Abstract：In industry, image contour information extraction is beneficial to the positioning of the workpiece in the later stage, and the effect of contour information extraction directly affects the accuracy of template matching. The gradient information of traditional Canny operator is less, and the applicability of setting double threshold value artificially is low. In this paper, adaptive median filter is used instead of Gaussian filter. Sobel operator 45°, 135° direction templates are added. The double threshold is calculated iteratively by using the maximum difference between classes. The results of workpiece show that the improved algorithm can extract lighter contour information more accurately and suppress false edges, and can improve the matching positiong accuracy.

Abstract： Aiming at the problem that different parts of human body present different structural characteristics in millimeter wave image, the gray difference between hidden objects and human body is small, and detection is difficult, a millimeter wave segmentation detection algorithm based on saliency vision is proposed. After bilateral filtering, the algorithm combined with OTSU and morphological operation to complete the pretreatment to obtain human body region, and then divided the region according to human body characteristics, using the symmetry decision strategy to determine whether the target exists in the region, and finally enhanced significance and K-means clustering segmentation algorithm to segment hidden items to complete detection. Experimental results show that compared with the typical active millimeter wave image detection algorithm, the detection rate of the proposed algorithm is improved by 8.28%, and the false positive rate is reduced by 1.58%, showing better detection performance.

Abstract：The industrial Internet platform is changing from single empowerment to comprehensive empowerment. The cloud and platform development of industrial software has brought new tools to the digital transformation of enterprises. By analyzing the technology and functional characteristics of industrial Internet platform cloud-based industrial software, this paper puts forward a functional view that reflects the logical architecture and advantages of cloud-based industrial software, helps industrial enterprises select and inspect software functions that meet business needs, and promotes enterprises digitalization. It also provides a corresponding reference for the product planning of software service providers.

Abstract：With the development of enterprise business and the continuous accumulation of business data, data types are uneven, and massive data is becoming more and more difficult to handle. The metadata system emerged as a platform to improve data quality and data analysis. Metadata is now at the heart of data warehouses and business intelligence, and vendors have come up with their own solutions for metadata systems. Existing solutions currently do not support the operation of updating business data. In view of the increase in business data faced by the current metadata system, the lack of rapid expansion support, the inability to process data in time, and the problem that metadata modification cannot be synchronized and updated to business data, resulting in metadata and business data being out of synchronization, this paper provides a distributed metadata solution that supports horizontal expansion of the system and supports reverse control of business data through metadata operations. Support business generation through metadata operations.

Abstract：To enhance the accuracy and performance of text similarity measurement of science and technology projects, this paper proposes a new approach for measuring text similarity of science and technology projects by combining TF-IDF and Simhash. Firstly, this method uses natural language processing technology to preprocess science and technology project texts to get a term set, then uses the TF-IDF method to calculate the TF-IDF value of each term in the term set, and selects the important term with higher TF-IDF value. Secondly, this method uses the Simhash algorithm to get the Simhash signature of the text through mapping the selected important terms into fixed binary strings. Finally, Hamming distance is used to calculate the similarity between two Simhash signatures. Experimental results show that compared to the traditional Simhash and TF-IDF, the proposed method can promote the evaluation metrics of precision, recall and F-measure.

Abstract： As an important part of the practical education system, experimental teaching is an important means to cultivate students' innovation ability and practical ability. Nowadays, there are some problems in computer hardware experimental teaching, such as the platform is large，the location of the experiment is fixed and the degree of laboratory sharing is low. Under this circumstance, a hardware platform of remote experiment system based on FPGA was proposed to address the above issues. The hardware platform uses FPGA to build the underlying structure, which realizes the remote configuration of the FPGA，and supports a wide range of experimental content. Experiments show that the hardware platform supports a wide range of experimental content. By using the hardware platform, schools can offer remote experimental courses with the same experience as those of offline experiments.

Abstract：Target detection is an important research content in the field of intelligent security. To meet the requirements of miniaturization and high real-time usage, a kind of real-time detection system is designed based on FPGA. The CMOS camera is used to collect video image. DDR3 SDRAM is taken as the cache medium of video image. Combined with the parallel processing characteristics of FPGA devices, the pipeline technology is used to modularize the inter-frame difference detection algorithm. The detection results are transmitted through Ethernet, which realizes image acquisition and motion object detection and display. Experimental test results show that the proposed system can detect moving targets quickly and accurately, with high stability and good real-time performance.

Abstract：Traditional experiments of free-flight in the wind tunnel use scale models to study aerodynamic parameters, with the increase in the type and number of collectors, the design of airborne equipment structure and control of data transmission are becoming more and more difficult. Aiming at this current status, a multi-channel data acquisition and transmission system based on FPGA logic control and timing drive is designed. The system adopts the three-terminal structure of airborne terminal, ground terminal and upper machine terminal, and the division of labor is clear. The airborne terminal adopts a miniaturized two-layer structured design to acquire the angle signal, wind direction sign signal, Inertial Measurement Unit (IMU) signal, aviation attitude signal, the ground terminal is responsible for the integration and distribution of data, and the upper computer terminal performs data monitoring and system control. After testing, the acquisition resolution of the system is ±0.07%FS(Full Scale) and a data transmission cycle is less than 2.42 ms, which is in line with the expected target.

Abstract： In this paper, a remote online software update method based on 1553B bus and Serial RapidIO（SRIO） high-speed serial port is proposed to solve the problems of large number of internal software, large scale, long update time, scattered to multiple modules, difficult disassembly and assembly of airborne equipment under the loading condition. The system composition and working principle are described, and the remote online update method and implementation process are given. This method effectively solves the online update requirements and software maintenance of complex stand-alone software with high reliability and high throughput under the condition of full ammunition. It has the advantages of flexibility, reliability and high data rate in practical application.

Abstract：In order to solve the problems such as high-bandwidth, multi-channel, large-capacity and high-integration in the data storage domain of advanced radars, a NVMe solid state disk storage array technology of multi-channel data based on FPGA is proposed. This storage unit integrated in the radar system is designed as a standard module. It takes FPGA as the core chip, which uses the methods of parallel double-group memory writing and reading data simultaneously, and dual-channel PCIe Gen3 processing data independently. And it designs complex logical function modules, achieves the transmission and processing for big data stream, and improves the storage bandwidth. This technology is applied to the radar in the industrial environment. The repeatable tests show that a storage unit can simultaneously store 16-channel data with the storage capacity up to 24 TB and writing bandwidth up to 7 GB/s, and can meet the system storage performance.

Abstract：In recent years, the localization of computer hardware and software systems has developed rapidly, and the general technical architecture and ecosystem Phytium Kylin Security (PKS) led by China Electronics has become more and more mature. At present, most of the research on the terminal performance scenario verification method is based on non-domestic system equipment, lack of pertinence, and it is difficult to fully explore the problems of domestic computer hardware and software system. This paper proposes a research on the terminal performance scenario verification method based on the PKS system, and conducts a comparative test of the terminal performance scenario verification method for common domestic equipment in China. The experimental results show that the terminal performance scenario verification based on the PKS system proposed in this paper can conduct more targeted and in-depth tests for domestic equipment, and at the same time meet the needs of domestic terminal performance scenario verification, and realize the systematic independent inspection of domestic equipment.

Abstract：With the development of communication technology and the increasing demand for smart campuses, traditional campus networks are also faced with major challenges such as insufficient bandwidth, inflexible networking, and difficulty in accessing the Internet of Things. In recent years, the industry has introduced network technologies such as PON and WDM in the campus network, or improved and optimized the existing Ethernet technology to deal with the problems and challenges. This paper summarizes the representative networking solutions in the industry, analyzes the technical principles, characteristics, and networking technology advantages and disadvantages, and uses the campus network scene as standard to judge and summarize the adaptability respectively, so as to achieve the comparison and judgment of technological advancement. Through the comparison, the shortcomings of the existing technology and the advantages and disadvantages of the new technology can be more deeply understood, in order to be used as a reference for solution selection in the future network construction.

Abstract：As a key component for offloading and accelerating key services in cloud data centers, smart NIC is significant for improving the processing efficiency of network and storage services in data centers. The service needs of data centers are complicated, lacking the support of mature commercial chips, and the design of smart NIC is difficult to meet the multiple requirements of service offloading flexibility, high performance, cost economy and so on in a limited space. Designing a separate architecture scheme to physically decouple the smart NIC interface function from the service offload processing function, break through the physical space limit of the network card, reuse the service offload processing hardware resources, and improve the cost economy is an effective way to achieve the engineering and large-scale deployment of smart NIC.

Abstract：PKS system is a representative of the autonomous computing systems of China. To promote the development of PKS system, it is important to construct PKS ecosystem. This paper analyzes the significance of PKS ecosystem construction, and then proposes a novel 5S model based on the experience of CEC Joint Innovation Research Institute, aiming to construct PKS ecosystem through five areas, including development strategy, organizational system, engineering service, resource sharing, and brand spreading. This model has been promoted and adopted in the institute with positive results, which is valuable and instructive for the future relevant work.