Search

Aiming at the demand for intelligent detection and control of alcoholic gas concentration in the fields of the national economy, people′s productions and lives and so on, a wireless intelligent alcohol concentration detector is designed and developed in this paper. The instrument adopts the STC90C52RC microcomputer to control the 10-bit A/D converter TLC1549 to acquire the output voltage signal of the alcohol concentration sensor TGS2620. The alcohol concentration value is obtained by data processing. This value is displayed in real time by the LCD module DM-162, broadcasted by the speech recording and playback circuit ISD2560, and sent to the upper computer by the wireless transceiver chip nRF24L01 respectively. The user interface of the upper computer is developed by employing the general object-based programming language Microsoft Visual Basic 6.0 to display and save the alcohol concentration data. In the software, the anti-pulse interference average filtering algorithm is used to suppress the pulsating interference in the sampled data. The piecewise linear interpolation method is used for scaling transformation so as to improve the real-time performance of the system. And the linear fitting technique based on least square method is used to correct the system error. The actual testing results show that the measurement accuracy of the instrument is less than ±2% and the communication distance is less than 100 m, which verifies the correctness and effectiveness of the design.

With the construction of BDS III satellite nearing completion, Beidou service will transition smoothly to be provided mainly by BDS III system. The B2b signal interface document was announced on December 27, 2019. B2b signals are divided into two types: B2b signal that provides basic navigation services and PPP-B2b signal that provides precise point positioning services. Based on the B2b signal interface control file, this paper introduces the B2b signal information structure, message type, and analysis results of satellite signals in the sky, and gives the positioning accuracy results of B2b signals. The PPP-B2b information in the sky is also analyzed. This article also explores the Q branch signal of B2b.

Artificial neural network(ANN) is used to extract scattering parameters and noise parameters of GaAs high electron mobility transistors with different frequency bands and gate widths. Based on the two neural networks, the two groups of scattering parameters and noise parameters are trained and studied respectively. The average relative error and mean square error are obtained by comparing different hidden layers and the number of neurons. It is found that 8-8-6 and 6-4 correspond to the optimal hidden layers and number of neurons of the neural networks with scattering parameters and noise parameters. The test results show that the average relative error of scattering parameters is 2.79%. Compared with the conventional single neural network structure, the average relative error is increased by 31.3%. This shows that the model in this paper has better accuracy and reliability, which shows that this model has higher accuracy and is very suitable for parameter extraction of RF transistors with wide band gap and strong nonlinearity.

With the establishment of the national emergency management department, state attaches more importance to emergency communications support after disasters. In order to solve the problem of the destruction of ground communications equipment after the disaster and the inability to guarantee normal communications, the research of quick“Sky to Ground Integrated” network building based on drone base stations has entered the planning and construction period. In order to meet the requirements of large bandwidth, low latency and low cost in the development of tethered type unmanned aerial vehicle(UAV) emergency system, based the key techniques such as large-scale antenna, edge computer and network slicing based on 5G, an emergency communication solution of tethered UAV based on 5G is put forward in this paper. The implementation architecture and key technologies of the scheme are analyzed, and the corresponding conclusion is given.

For DC off-grid photovoltaic power generation system, when the power demand of DC bus is large, Maximum Power Point Tracing(MPPT)is needed to keep the solar panels in the state of maximum power output for the output voltage and power are affected by temperature and illumination evidently. When the power demand is small, all converters working at the same time will lead to low efficiency of light load. The most simple and direct way to improve is to reduce the number of working modules, so as to ensure that the power system can achieve energy conversion efficiently within the full range. In order to solve the above problems, this paper proposes a control strategy for power generation system. This strategy makes full use of the solar array and adjusts the number of working modules dynamically according to the load demand. It provides a rationale for improving the operation efficiency of DC off-grid photovoltaic power generation system within the full load range. Finally, the effectiveness of the scheme is verified by simulations.

Based on the characteristics of suddenness, complexity, dynamics and lack of information of navigation accidents, UAV technology has been organically integrated with China′s navigation accident emergency rescue system, becoming a research hotspot of maritime emergency communication. The user-centered UAV base station group network breaks the traditional base-centered architecture, and selects the set of potential service sites for each user to build a dynamic base station group, so that users can make full use of the diversified gain of potential service sites within the base station group and the cooperation gain between potential service sites. In this paper, a user-centered base station group networking scheme for UAV is proposed, and the user initial access scheme suitable for ship formation scenario is designed and its performance is simulated and evaluated.

Radio spectrum occupancy prediction is a key technology in cognitive radio research. In the traditional method, only single step size prediction can be performed, and the effect of multi-step size prediction is significantly reduced. Referring to the idea of searching frequent item sets in Apriori algorithm，a multi-step spectrum occupation prediction method based on association rule mining is proposed. In terms of data acquisition, the electromagnetic spectrum detection system developed by the unit is used to monitor the frequency spectrum of FM broadcasting service(88~108 MHz) for 48 hours continuously, and some data whose occupancy meets the demand of spectrum occupancy prediction are selected for prediction and analysis. The experimental results show that the proposed method is effective in FM broadcasting frequency band, and the prediction accuracy of multi-step length is more than 70%. This paper also briefly analyzes the important parameters of the algorithm and points out the influence of the parameters on the accuracy of the output results.

The paper firstly analyzes the formation mechanism of signal reflection in signal integrity, lists several typical waveforms of signal reflection ,and then puts forward the main measures to eliminate signal reflection. It focuses on the analysis of termination matching and the effective measures to be taken under different topologies, and gives the simulation verification of signal integrity combined with engineering cases.

Emergency communication support requires rapid deployment, quick on-line and simplified configuration process, but the network architecture based on 3GPP standard is too large and complex to meet. Lightweight 5GC aims at referring to the standard architecture, carrying virtualization technology based on the common hardware platform, realizing the decoupling of hardware and software, controlling and forwarding the cloud core network, and reducing the business process and improving the business processing capacity through the integration and tailoring of network element functions, becoming a new solution for the emergency communication industry.Based on the application scenario of emergency communication private network, this paper analyzes the implementation architecture and key technologies of lightweight 5G core network, and explores the deployment scheme based on typical cases of emergency support.

In order to adapt to the needs of rapid deployment and convenient configuration in 5G emergency communication scene, the terminal, base station, user plane function(UPF), mobile edge computing(MEC), core control and other network devices must be simple, efficient and flexible. In order to quickly build emergency communication private network application scenario, this paper puts forward a scheme of integrated edge fusion technology and equipment implementation，which provides a highly integrated wireless access，data forwarding and edge application service ability for 5G field emergency communication，at the same time, also simplifies the complexity of equipment operations and external network requirements, and improves the service performance of the whole system.

As an important supplement to the daily communication system, the emergency support communication system plays a vital role in the event of a disaster. Aiming at the requirements of the emergency support system in the event of a disaster, this paper proposes a technical solution based on 5G network slicing. Based on the flexible network customization capabilities of 5G, the performance requirements of various scenarios of the emergency support communication system on the network are analyzed. Based on the typical emergency support communication system architecture, the implementation process of the 5G network slicing of the emergency support system is discussed to realize the priority guarantee of 5G network resources for emergency services, ensure the real-time interaction and sharing of information, and improve the scientificity and timeliness of information decision-making.

In the construction industry, the cement floor is often polished by manual and mechanical methods. The working environment is harsh and the working efficiency is low. The automatic polishing robot can replace manual operation and improve the working efficiency. In this paper, the basic motion of the cement floor polishing robot is studied, and the motion control scheme of the robot is proposed. Firstly, the kinematics model of the robot and the speed control algorithm to realize the basic motion were designed. Then, the design of the motion control system of the cement ground grinding robot was designed with STM32 single chip microcomputer as the control core, and the basic motion control in the complex trajectory of the robot was realized. Experimental results show that the maximum positioning error of the robot is 3.1 cm when it moves in a straight line and turns. The reliability and stability of the system for the basic motion control of the robot are verified, which provides a reference for the control of similar robots.

The implementation of high-performance Viterbi decoding algorithm in resource-constrained processors is a hot topic in recent years. This paper is based on the XC6SLX16-2CSG324 type FPGA processor. In the case of limited resources, in order to balance the problem of Viterbi decoding delay and resource consumption, an improved algorithm is proposed. On the basis of the traditional Viterbi decoding algorithm, the purpose of controlling the path metric value is achieved by maximizing the pre-defined storage path metric value register, and then the stepped survivor path information storage structure is used to complete the storage of the surviving path information and simplifying the decoder hardware implements complexity, reducing decoding delay and resource consumption. Based on the ISE Design Suite 14.7 platform, FPGA-based verification of(2，1，4)-dimensional bit decoders with 20- and 3-bit soft-decision depths is performed, and combined with MATLAB simulation. The results show that the method can effectively reduce the decoding delay and reduce the resource consumption.

In this paper, according to the requirement of safety design of IoT(Internet of Things) operating system, combined with the research and engineering practice of classical linear temporal logic, logic programming and theory of formal model detection, a methodology for safety design of IoT operating system is proposed, and the engineering prototype is validated. Practice has proved that the effect of this method is in line with expectations. It is not only suitable for the safety design of IoT operating system, but also can be further extended to other areas of software product design with high safety requirements.

The traditional method for feature extraction contains limited discriminant features, and the deep learning method need lots of labeled data and it′s time-consuming. This paper presents a method which fuses the deep and shallow features for face recognition. Firstly, the HOG feature is extracted from each images and the dimensionality reduction is followed; and the PCANet feature is extracted simultaneously and its′ dimension is reduced. Secondly, the fusion of the two types of features is conducted and discriminant features are extracted further. Finally, the SVM is adopted for classification. Experiments on the AR database verify the effectiveness and robustness of the proposed method.

Country attaches great importance to the planning and construction of emergency communication network, especially with more attention paid to the UAV(Unmanned Aerial Vehicle) emergency communication technology research. Relative to the 4G network characteristics of the "Try best", 5G with large bandwidth, low latency and high reliability of network capacity,provides a flexible configuration space for performance of a network configuration, and will gradually become the preferred technology meet the demand of emergency communication different scenarios. UAV is proposed in this paper, based on fixed wing type 5G of emergency communication coverage solution. The solution achieves network through the integration of UAV airborne 5G base stations, combined with ground acer station auxiliary. UAV base station downlink coverage and back link budget are simulated，the results show that the proposed solution can meet the demand of remote areas and emergency communication coverage.

With the increasing demand for power distribution and control in new energy generation, hybrid vehicles, and industrial equipment, the operational reliability of IGBTs needs to be addressed. This paper combs the main failure mechanism of IGBT module, and based on this theory, summarizes the research on active thermal control of IGBT module at home and abroad. Finally, the future of active thermal control is expected to be combined with the actual situation.

The underwater acoustic monitoring acquires a wide range of frequency bands, including multi-band feature information in the recording process, and doped with various kinds of noise. In order to extract the effective feature information of each frequency band of the monitored signal reliably, avoid the influence of noise, solve the drawbacks of energy method in judging the characteristic frequency band, and break the limitation of single method in extracting multi-band feature information of broadband signal，in this paper, using the advantages of wavelet packet, correlation coefficient and Hilbert-Huang，multi-band feature information of broadband signals is extraded. Firstly, the fine frequency division of broadband signal is realized by wavelet packet algorithm, and the validity of node signal is judged and the noise nodes are eliminated by correlation coefficient analysis. Secondly, the effective components and noise in the effective node band are further separated by Hilbert-Huang. Finally, the effective band characteristics are obtained by reconstructing the node coefficients after processing.The results of simulation and experimental signal processing show that this method has certain advantages in extracting characteristic frequency information of broadband underwater acoustic monitoring signals, and can be popularized and applied in underwater acoustic target recognition and monitoring.

This paper is aimed at researching and designing LDOs without off-chip capacitors. Whether the LDO can stabilize the output voltage when there is no external capacitor is studied. Based on this, a new type of high-stability LDO structure with no off-chip capacitor is studied to apply power to the internal circuit module in the DC-DC converter. The poles of the LDO are adjustd by adjusting the internal op amp structure and Miller compensation of the op amp. The simulation results show that the structure has high stability and can output a stable voltage.

Network slicing is one of the hot technologies in 5G, and the flexibility and timeliness of slicing management are important issues in deployment. Based on the introduction of the creation and management mechanism of 5G network slices, this paper proposes a network slice management architecture based on the openness of 5G network capabilities, and discusses the optimization process of network slice parameters and typical application scenarios.The timeliness advantage of 5G capability opening will greatly enhance the flexibility of network slice management and ensure the user′s business experience.

State Grid Corporation is comprehensively promoting the construction of ubiquitous power Internet of Things. Power wireless private network is the important wireless access network. The research and application of new intelligent terminal is the important construction content. According to the business needs of ubiquitous power Internet of Things, this paper developed the power wireless private network telecommunication terminals. The functional design and development were introduced respectively from the system architecture, hardware, and software. The key design and parameters were tested to verify the performance. The power user electric data acquire business was selected to test the service adaptation and business function of the terminal. The results showed that the telecommunication terminal can meet the requirements of the standard specification, and can bear the power user electric data acquire business.

This paper proposes to apply Transformer model in the field of Chinese text automatic proofreading. Transformer model is different from traditional Seq2Seq model based on probability, statistics, rules or BiLSTM. This deep learning model improves the overall structure of Seq2Seq model to achieve automatic proofreading of Chinese text. By comparing different models with public data sets and using accuracy, recall rate and F1 value as evaluation indexes, the experimental results show that Transformer model has greatly improved proofreading performance compared with other models.

The space solar array simulator(SSAS) simulates the output characteristic curve of the solar panel under various load conditions, which is used to complete the performance test of the tested power system, and is widely used in the field test field of the spacecraft power system. However, in the actual test environment, the long test cable between the solar array simulator and the power supply of the spacecraft being tested seriously affects the simulation accuracy of the output curve of the solar array simulator. Therefore, the compensation connection line is generally used to remotely compensate the output voltage of the simulator. This paper proposes a simple method to detect the connection state of the remote voltage compensation terminal connection line, and gives a normalized fault state judgment standard. This method is not only suitable for solar array simulator equipment, but also for any high-power power supply where remote voltage compensation is required for the output voltage.

Aiming at the long broadcast, much fan in/fan out data path problem brought by high parullelism in the process of the Field Programmable Gate Array(FPGA) to realize the convolution computation in convolutional neural network, this paper adopts pulse array to realize convolution calculation module of convolutional neural network, fixes weights to each processing unit, according to the dimension of the input and output characteristic figure sets to pulse array size, and finally by Vivado high level synthesis realizes convolution calculation module hardware design. The experimental results show that the design has low resource occupancy and good expansibility while realizing the time-series requirements of level 1 pipelining.

This paper presents a rail-to-rail constant transconductance operational amplifier input stage. The proposed circuit changes the tail current of the input differential pairs dynamically for a constant-gm by using dummy input differential pairs. The problem which causes total gm variation is input pairs and dummy input pairs can not take effect at the same time with the common-mode input voltage changes, because the tail current transistor of the input pairs are in triode region when the input pairs are turned off, the dummy input pairs will enter subthreshold region from cut-off region before the input pairs when common-mode voltage changes. The effect of this problem is more obviously in low supply voltage design. To solve this problem, compensate current sources is added to the tail current transistors of each dummy input differential pairs for lower gm variation. The gm of this Op Amp′s input stage varies around ±2%.

In order to promote the power, economy and safety of new energy vehicles and estimate accurately the state of charge(SOC) of lithium batteries for vehicles, this paper takes pure electric vehicle power lithium batteries as the research object, uses genetic algorithm(GA) to optimize BP neural network, and handles the shortcomings of slow convergence speed, weak global search ability and easy to fall into local minimum of error back propagation. A neural network model for predicting SOC value based on GA-BP algorithm is presented. Compared with traditional BP algorithm, the simulation outcomes show that the algorithm takes into account the learning speed, small error, global search ability of the neural network and meets the requirements of estimating SOC value of power battery.

To make power electronic devices miniaturization and lightweight, GaN FET is paid extensive attention in power electronic application fields because of its advanced performance than Si device. GaN FET structure and its driving circuit have important impacts on operating safely and stably. In the paper, firstly, the GaN FET structure and its products are introduced; Secondly, the isolated method is described, then the principles of the discrete driving circuit and the integrated driving circuit are presented; Finally, GaN FET′s application is summarized.

With the rapid development of the Internet, each platform often uses metadata to organize and manage system data resources. The system within the platform is closed so that metadata resources are dispersed and stored repeatedly, which increases the difficulty of metadata sharing and reduces the utilization of metadata. In order to solve this problem, this paper proposes a public metadata configuration management platform, which provides unified configuration and management of public metadata, provides configuration and shared metadata services by using Web Service technology, and improves data sharing efficiency by using Redis memory database. At present, a configuration management platform based on meteorological public metadata has been used in the meteorological big data platform, which not only improves the utilization of the platform common metadata, saves storage, but also improves the response speed of the service. This article abstracts the main functions of the configuration management platform, making the system design applicable to other similar metadata-based data sharing platforms.

5G network slicing has received widespread attention in the vehicular network industry, and the development and evolution of various scenarios of vehicle-road collaboration also reflect that 5G applications are gradually deepening. In this paper, the development process of vehicle-road collaboration and the network performance requirements are analyzed firstly. Then we present the system architecture of vehicle-road collaboration. Subsequently, implementation cases and specific processes are given out, including network slicing creation process, scheduling system for slice resources. Finally, the key technologies of network slicing to ensure the vehicle-road collaboration security are introduced.

The power system promotes the construction of a strong smart grid and a ubiquitous power Internet of Things, making full use of modern information and communication technologies to support the evolution of traditional power grids to the energy Internet. In recent years, network slicing has received widespread attention from researchers, and it is not only a technology but also a new business model, which has attracted broad interest of researchers. Based on above background, in this paper, several typical user cases for smart grid are introduced firstly. In addition, we present the E2E business models and steps to implement network slicing for operators. Besides, the Total Cost of Ownership(TCO) and Return on Investment(ROI) of Network Slicing are also analyzed in Smart Grids for Operators.

The received signal amplitude of zero-IF Doppler radar usually has a large dynamic range, and this large dynamic range brings inconvenience to signal acquisition. In view of the large dynamic range of this received signal, this paper designs a signal conditioning circuit from the perspective of negative feedback, which includes low noise amplification module, automatic gain control module, non-diode precision rectifier module and subtraction amplifier circuit module; it realizes the function of automatic gain control. After the computer software is simulated, the hardware circuit is fabricated, and the function of the debugging circuit meets the requirements.

In order to solve the problems of the efficient synergy between the management orchestration components required for the deployment of the edge computing business and the NFV MANO components in the telecommunications cloud, this paper firstly analyzes the division of responsibilities and interaction processes between the management orchestration components of edge computing and telecommunications cloud. Secondly, based on the deep learning method and model, a solution is given to improve the management orchestration efficiency between the virtual network elements of the telecommunications cloud and the edge computing applications. Finally, the main issues and directions in this area are prospected.

In the space rendezvous and docking, real-time measurement of the position and attitude between the visual sensor on the tracking spacecraft and the characteristic target on the target spacecraft is needed, and the premise is to quickly and accurately identify the feature targets. In this paper, a fast recongnition algorithm of four concentric ring feature targets is proposed. Firstly, the algorithm uses the optimized Otsu for image segmentation. Then the contour is extracted by the connected region recognition method. Finally, according to the relationship between the circumference and the area of the circle, the circle is determined, and the least square method is used to fit the circle and extract the center of the feature circle. The experimental results show that the method can recognize feature targets accurately and quickly within 2 m distance, and the computational efficiency to the method is improved by nearly 10 times compared with the pre-optimization algorithm.

Crop yield forecasting plays a decisive role in the development of national economy planned by the government. It is of great significance to rationally plan planting strategies and reduce the waste of water and fertilizer. There are many factors affecting crop yield, and accurate prediction of crop growth means a lot. Climate is an important factor affecting crop yield. Based on climatic factors, this paper proposes a time series prediction method based on the long short-term memory(LSTM) for crop yield prediction. The method combines historical yields with climatic factors to forecast the crop yield of the next period by years. Experimental results show that compared with LSTM and support vector machine methods, the proposed method achieves higher accuracy in crop yield prediction.

With the rapid development of 5G technology, the number and density of 5G base stations will be far higher than 4G, and the construction and maintenance of base stations will become a problem that cannot be ignored. Therefore, this paper comprehensively analyzes the 5G base station decommission situation and puts forward a 5G base station decommission cost estimation scheme based on big data. With the base station historical decommission data, the LSTM neural network is used to establish the base station decommission prediction model. Then the 5G base station decommission cost estimation model is constructed, and the cost estimation of the forecast 5G base station decommission is made. Finally, through the experimental analysis, the effectiveness of the scheme is explained and suggestions are put forward.

The existing FFT-WT algorithm and FFT-BP algorithm have advantages only for the detection of certain specific harmonics in photovoltaic systems. In this paper, the FFT-WT algorithm is improved, and a DFFT-WT algorithm is proposed. The FFT-BP algorithm is improved and an improved FFT-BP model is proposed. On the basis of this, the two are combined, the threshold interval is introduced to limit the search scope of the neural network iteration, and a DFFT-WT-BP(Double FFT-WT-BP) detection algorithm is proposed. The simulation results show that the algorithm can detect complex harmonic signals composed of multiple harmonics in the actual photovoltaic network inversion system, and is more accurate and practical.

In response to the national call for energy conservation and emission reduction, the energy saving and consumption reduction of base stations has become the top priority in today′s communications field, the energy-saving technologies of major operators will also become core competitiveness. This paper studies the Artificial Intelligence(AI)-based 5G base station energy-saving technology. Traditional base station energy-saving technologies are analyzed. The prospect of 5G base station energy-saving technology combined with AI technology is explored. The development direction of AI energy-saving technology has been deeply studied. The AI energy saving model and the AI-based collaborative energy saving scheme are highlighted.

Filter Bank Multi-Carrier(FBMC) is a modulation scheme that may replace OFDM in 5G mobile communications, but its peak to average power ratio(PAPR) is higher. The DFT-spreading technique is to spread the input signal with DFT, which can reduce the PAPR of FBMC signal to the level of single carrier transmission. But there are still shortcomings: after derivation, the PAPR is substantially high due to the four different symbols added at the symbol timings in the transmit waveform x(t). For this shortcoming, this paper improves the phase shift terms and proposes an identically time shifted multicarrier-DFT spreading(ITSM-DFTs) algorithm, which fully exploits the single carrier effect of DFT-spreading. Theoretical analysis and simulation results show that the algorithm effectively reduces the PAPR of the FBMC system.

With the advent of 5G era, a variety of application scenarios and differentiated service requirements are emerging, which puts forward higher requirements for the ability of operators to schedule network resources and service requirements. Based on the analysis of the new features of 5G network, this paper proposes an intent-based network(IBN) management scheduling system. Combined with the network architecture of IBN, various forms of service intentions are collected and translated into the required network operation and configuration, so as to realize the rapid and automatic deployment of the network. A fast scheduling algorithm is proposed based on this system to effectively schedule tasks at multiple levels and peer levels. The simulation experiments show that the fast scheduling system based on IBN improves the task processing efficiency and reduces the task scheduling processing time, which is of great significance to the service requirement scheduling guarantee of communication scenarios such as major events.

Digital Rights Management(DRM) is an information security technology that protects the rights and interests of publishers of digital publications. It guarantees authorized users to use digital information and protects the interests of copyright owners. However, the traditional central copyright protection method has some problems such as low security and rough authority control, which makes piracy and infringement occur frequently. Blockchain has the characteristics of decentralization, tamper-proof, public and transparency, which provides a new idea to solve the problem of copyright protection. This paper analyzes and compares the traditional digital copyright protection method and the copyright protection method based on blockchain, and puts forward the prospect for the future.

As the scale and complexity of integrated circuits increase, the verification work becomes more and more complex and important. The verification cycle has reached or exceeded 70% of the entire chip design cycle. Therefore, it is urgent to find an efficient verification method to improve verification efficiency and enhance reusability of the verification platform. The UVM verification methodology based on the SystemVerilog language can effectively improve the verification efficiency and shorten the verification cycle. Therefore, this paper uses the high-level abstract model C_Model as the reference model to access the UVM platform, verify the encoding module of the tag transmission link in the digital baseband processing unit, design random and non-random testcase, and send it through the driver and monitor verification components. Monitor and collect data, including the data generated by the hardware design RTL code and the data generated by the reference model, and then send the two data to the designed UVM scoreboard module for comparison, thereby verifying the function of the RTL and verifying the pros and cons of the system. It can be reflected by functional coverage. The verification results show that the alignment is correct and the functional coverage reaches 100% in the UVM scoreboard.

Parameter prediction of insulated gate bipolar transistor(IGBT) can effectively avoid the economic loss and safety problem caused by its failure. Based on the analysis of IGBT parameters, the paper designs a SoC hardware system of IGBT parameters prediction based on LSTM network. The system uses ARM processor as the general controller to control the call of each sub-module and the transmission of data. In the FPGA, the algorithm of matrix vector inner product is optimized to improve the data operation speed in the LSTM network.And the polynomial approximation method reduces the resources occupied by the activation function. The experimental results show that the average prediction accuracy of the system is 92.6%, the calculation speed is 3.74 times faster than the CPU, and the system has the characteristics of low power consumption.

A method for control of three-phase unbalanced load based on variable frequency power was proposed in this paper. The general idea was introduced firstly, and then the implementation of the scheme and the design method of variable frequency power supply and the realization scheme of the master control center at the side of low-voltage distribution network was described in details. When three-phase unbalance occurs in the power grid, the general control center sends phase-changing instructions, and the inverter power supply receives instructions to perform phase-changing, and the normal work of the load is not affected in the phase-changing process, which can effectively solve the problem of three-phase unbalance in the power grid. By setting up the basic model of low-voltage side commutation with variable frequency power as the core, the rationality of the commutation process is simulated and analyzed by MATLAB, and finally the feasibility of three-phase unbalanced control scheme with frequency conversion power as the core is verified by experiments.

In recent years, convolutional neural networks(CNN) and recurrent neural networks(RNN) have been widely used in the field of text classification. In this paper, a model of CNN and long short term memory network(LSTM) feature fusion is proposed. Long-term dependence is obtained by replacing the LSTM as a pooling layer, so as to construct a joint CNN and RNN framework to overcome the single convolutional nerve. The network ignores the problem of semantic and grammatical information in the context of words. The proposed method plays an important role in reducing the number of parameters and taking into account the global characteristics of text sequences. The experimental results show that we can achieve the same level of classification performance through a smaller framework, and it can surpass several other methods of the same type in terms of accuracy.

In this paper, trusted networking is proposed as one of the most important technologies trend in 6G network according to the status of services and applications in 5G network. In 6G network, network trust is proposed as a complementary concept of network security. Compared with network security based on the presumption of guilt, network trust is designed based on the presumption of innocence to prove itself without network attack of threat. Finally, trust requirements in next generation network are analyzed with the trend of higher frequency spectrum, capital expenditure, operational expenditure, and multi-domain integrated, including random access network sharing, space-ground integrated network and massive connection Internet of Things.

The planar integration design of the antenna system for soldier terminals is presented in this paper based on the requirements of the miniaturization and high integration. A four-antenna system including recieving and transmitting Ad-Hoc network antennas, satellite navigation and communication antennas is proposed to solve the 3D packaging and low-integration problems. The conventional 3D Ad-Hoc network antennas and circular polarized helix antennas are replaced by planar printed monopoles and horizontal dipoles with linear polarization, respectively. Simulated and Measured results show that the planar four-antenna system realizes onmidirectional radiation in horizontal plane in Ad-Hoc network frequency band and upper space coverage in satellite navigation and communication band. This antenna system can be applied in soldier terminals to improve its integration.

This paper derives a state augmentation-based modified iterated extended Kalman filter(SMIEKF) by making use of the state augmentation technique, in which the state variables are directly augmented with the measurement noise. The improved performance is then demonstrated by a target tracking problem involving radar measurements of an atmospheric re-entry vehicle. Simulation results clearly show that the proposed filtering algorithm exhibits faster convergence rate and better estimate accuracy compared to the modified iterated extended Kalman filter(MIEKF) and traditional extended Kalman filter(EKF).

This paper presented the design of a novel general signal processing system based on the partial reconfiguration of Field Programmable Gate Array(FPGA). This solution uses framework of high speed FPGA，Digital Signal Processor(DSP) and Analog/Digital Convertor(DAC), and supports the performing, storing and dynamically reconfiguring of dozens of waveforms in three channels. This system is suitable for the miniaturization application.

The chip emulator is an effective debugging tool in the embedded software development, and many existing chip emulators lack the protection of the software program during the debugging process. This paper proposes a solution to control the access rights of software programs during the debugging phase. The solution mainly includes three steps: firstly, the program access rights is generated in the form of an encrypted configuration file, then the program data is encrypted, and finally the encrypted configuration file and the encrypted program data are downloaded to the chip emulator. The solution realizes flexible configuration of program access rights and adopts encrypted files to prevent unauthorized tampering of configuration rights and program data, and increases the security of debugging.

In 5G mobile communications, key technologies such as large-scale array antennas, new scenarios, new services and new frequency bands are introduced. The goal of future network evolution is to achieve independent network decision-making and independent evolution through the introduction of artificial intelligence. 5G network planning also faces the requirements for intelligent evolution. From the perspective of 5G network planning and deployment phases, this paper discusses how to build an overall 5G network planning solution based on big data, machine learning and other technologies to guide the planning and construction of 5G network planning.