Abstract：With the commercial deployment and technological evolution of 5G networks, mobile communication networks are facing the coexistence of 2G/3G/4G/5G multi-mode networks, and the network structure is becoming increasingly complex. The distinctive features of 5G networks with multiple networking modes and multiple terminal categories further aggravate the complexity of network topology and boundary scenarios. How to ensure user′s mobility perception has become a crucial part that affects the reputation of the networks. This paper conducts in-depth research based on the solutions of 5G network mobile performance optimization, and introduces in detail the interoperability strategies of different terminals in different networking modes, so as to provide a theoretical basis for optimization of 5G network mobility and cooperative networking, and effectively promote 5G user′s perception improvement.

Abstract：In the third year of 5G construction in 2021,the focus will be on the scale deployment of 5G indoor distribution system.The characteristics of low cost and low failure rate of passive indoor subsystem conform to the national construction and sharing strategy.Intersystem modulation interference has become an urgent problem to be solved in the construction of multi-system integration.Based on the in-depth analysis of the principle of IMI, this paper studied the possible IMI of 5G cell subband, proposed the solution of IMI of 5G cell subsystem, and combined with the case of a cell submodulation interference of Tower Company, put forward suggestions on how to solve the problem of IMI.

Abstract：When building a large-scale 5G fronthaul network, operators have deployed a large number of CWDM passive coarse wavelength division multiplexing(CWDM）equipment in order to save fiber resources to realize 3-channel fronthaul transmission.In order to meet the needs of 5G multi-frequency networking and the joint construction and sharing of China Telecom and China Unicom, operators have proposed multi-channel solutions including medium wavelength division multiplexing(MWDM), LAN wavelength division multiplexing(LWDM), dense wavelength division multiplexing(DWDM) and other technologies.In this context, the research team innovatively proposed a multi-channel solution based on CWDM/circulator technology. This solution not only achieves industry chain sharing, has the highest localization capability, but also minimizes 5G fronthaul network construction cost. This article describes in detail the current network deployment plan of the CWDM/circulator solution and the existing problems.

Abstract：With the continuous development of mobile communication network, operators have different frequency band resources at high, middle and low levels. Radio frequency band resources are valuable strategic resources, and how to use spectrum resources reasonably and efficiently is the primary issue that operators consider at present. Based on the comparative analysis of the coverage performance and service bearing of 5G NR 2.1 GHz and NR 3.5 GHz bands, this paper puts forward suggestions for networking and inter-carrier mobile strategy in different scenarios, providing an important basis for the subsequent construction of 5G networks.

Abstract：When 5G mobile communication technology is applied to sea coverage, it is not only to solve the problem of difficult station construction, but also to solve the parameter configuration and coverage distance estimation under ultra-far sea coverage. In this paper, firstly, the channel model of sea surface ultra-far coverage is built, and the equivalent earth model is given considering the influence of air permittivity change on radio wave propagation. Secondly, several key parameters affecting cell coverage radius in 5G are analyzed in detail. Finally, the main frequency bands of the three operators are simulated, and the relationship curves between the maximum coverage distance and the path loss in different frequency bands are given.

Abstract：As an important field of the development of new generation information technology, artificial intelligence and blockchain have been regarded as the national strategic frontier technology. With the development of artificial intelligence and blockchain technology, while constantly breaking through the technical development in their own fields and promoting industrial transformation and upgrading, they are also facing the problems of data island, high technical barriers and high computing cost. Combined with the characteristics of artificial intelligence and blockchain technology, this paper analyzes the six key points of integration, such as data sharing, algorithm sharing and computing power supply, and preliminarily explores the technical problems that can be solved by the integration of the two, so as to provide reference for promoting the development of new generation information technology to diversification, intelligence and security.

Abstract：Considering that the current power industry still lacks effective domain word discovery methods, this paper takes the power industry science and technology project text as the original corpus, combines the statistical features based on the mutual information, left entropy as well as right entropy with the features of traditional language word-formation rules, and proposes the new concept of power text word formation rate. The proposed method firstly uses the word formation rate to get the initial candidate word set by unsupervised filtering, and then performs the text slicing algorithm and common word filtering operation on the candidate word set, and finally performs the word embedding and spectral clustering algorithms to get the final power text-domain words. Experimental results show that the method proposed in this paper is accurate and effective, and provides a new method for power text domain word discovery.

Abstract：Object detection algorithms based on deep learning are difficult to deploy on low computing power platforms such as mobile devices due to their complexity and computational demands. In order to reduce the scale of the model, this paper proposed a lightweight object detection algorithm. Based on the top-down feature fusion, the algorithm built a feature pyramid network by adding an attention mechanism to achieve more fine-grained feature expression capabilities. The proposed model took an image with a resolution of 320×320 as input and had only 0.72 B FLOPs, achieved 74.2% mAP on the VOC dataset and the accuracy is similar to traditional one-stage object detection algorithms. Experimental data shows that the algorithm significantly reduces the computational complexity of the model, maintains the accuracy, and is more suitable for object detection with low computing power.

Abstract：The application of SRAM memory in SoC is more and more common, and the design of radiation hardened memory has become one of the most important issues in the design of radiation resistant SoC. In this paper, two kinds of SRAM read circuits are proposed and compared in terms of readout circuit structure, data readout speed and single event upset resistance. The SRAM of the two readout structures have good single event resistance, but SRAM with dual-input DICE structure has shorter readout time compared with SRAM with single-input DICE structure.

Abstract：Aiming at the timing requirements of DDR3 system, timing simulation and optimization were carried out for DDR3 package and substrate design in a SiP(System in Package). Through simulation guidance design, the design success rate of DDR3 in SiP product was improved and the design cycle was reduced. The signal scattering parameters were extracted by ANSYS SIwave software, and then the topology construction and timing simulation analysis was carried out through Cadence SystemSI software. The relationship between signal timing and waveform was discussed based on the theory of signal integrity. The actual optimization scheme was given by combining with layout analysis. Finally, the designed DDR3 system could meet the timing requirements of JEDEC protocol through simulation iteration verification.

Abstract：In order to adapt to the application scenarios of low power consumption in the Internet of Things, and meet the requirements of low power supply and low input common-mode voltage, this paper proposes a dynamic comparator, with dual-positive feedback loop, suitable for ultra-low input common-mode voltage. The comparator uses a timing switch to control the input and output, which solves the problem that the traditional dynamic comparator cannot work properly when the input voltage is lower than the threshold voltage, and increases the input dynamic range. Only two MOS devices are stacked in series between the power supply and the ground, which reduces the minimum power supply voltage. Two positive feedback loops are introduced to improve the resolution. TSMC 180 nm CMOS process is used to design and verify the proposed comparator. The simulation results show that the lowest common mode voltage of the proposed comparator is 51 mV when the power supply voltage is 900 mV and the differential mode voltage is 1 mV, which is 374 mV and 264 mV lower than the traditional StrongARM and DoubleTail dynamic comparators, respectively. When the input common-mode voltage is lower than the threshold voltage, it achieves best delay among three topologies at moderate power consumption.

Abstract：High voltage multiplexer is an important part of battery management system. The high voltage multiplexer studied at present will introduce leakage current to the battery, which will affect the accuracy of battery voltage detection. To reduce the influence of the control signal of the high voltage multiplexer on the leakage current of the battery and improve the accuracy of voltage detection, a new type of high voltage switch with gate drive was designed. The high voltage switch adopts gate bootstrap structure, can be directly cascade, and the switch control signal does not introduce leakage current to the signal source. At last, a 7-channel high voltage multiplexer was built by using 0.5 μm 40 V BCD process to carry out simulation verification. The post-simulation results show that the isolation of the adjacent channel of the high voltage switch is -47 dB@1 MHz, the maximum input voltage is 35 V, and the circuit layout area is 1.2 mm×1.5 mm, which meets the requirements of application.

Abstract：Aiming at the teaching experiment of undergraduate or graduate students in power electronics, based on LabVIEW and simple FPGA board, a high cost-effective hardware in the loop real-time simulation teaching experiment platform for power electronics is designed. The simulation platform system is composed of upper computer and FPGA board. The man-machine interface and experimental waveform data display are based on LabVIEW software platform. Data exchange with FPGA board is realized through Ethernet. The platform system can download and install various circuit topology models and HIL real-time simulation experiments.

Abstract：Aiming at unsatisfied ideal accuracy of electronic nose while testing the concentration of gas pollutants，the particle swarm optimization and artificial bee colony algorithm based extreme learning machine(PSOABC-ELM) algorithm is proposed. The accuracy of electronic nose concentration detection is enhanced by improve extreme learning machine weights of input layer and hidden layer and hidden layer threshold random defects. PSOABC-ELM is compared with other algorithms and validated on the public data set. The results show that the PSOABC-ELM algorithm perform better than the others when Testinggas concentration of electronic nose, and the detection result error is smaller and the algorithm stability is stronger, which provides a new method for the detection of gas concentration of electronic nose.

Abstract：This paper introduces the non-public network technology put forward in 3GPP R16 and two networking methods as well as application status. It analyzes the enterprises′ demand for 5G private network and the trend of integration of NPN and public land mobile(a public land mobile network，PLMN). Based on the relationship between NPN and PLMN, two typical 5G public-private integrated deployment modes are proposed: continuous mode and overlapping mode. In order to solve two typical problems in these two modes: service continuity and connection connection, two technical solutions of R17 are discussed.

Abstract：For the faster-than-Nyquist(FTN) system, a joint convolutional coded modulation scheme with non-orthogonal shaping pulses, Gaussian pulse and extended Gaussian functions(EGF), is proposed and compared with the traditional FTN with the orthogonal waveform and the traditional orthogonal Nyquist systems. On the condition of approximate equivalent 99.99% power bandwidth and 5 iterations, the simulation results show that compared with the FTN system based on the raised cosine pulse, the FTN systems with EGF and Gaussian pulse can get 4.4 dB shaping gain and have only one sixteenth equalization complexity due to good time-frequency gathering feature.

Abstract：A Pearson correlation coefficient virtual machine selection approach called PC-VMS was proposed and discussed in this paper. PC-VMS uses the calculation method in statistics of Pearson correlation coefficient for historical CPU utilization data of virtual machines, and establishes a measurement of the CPU utilization of each pair of virtual machines. The mathematical model of the correlation between the rates was also constructed. The PC-VMS algorithm will obtain the CPU utilization of the last n times for each pair of virtual machines, calculate the Pearson correlation coefficient based on the two sets of input data, and finally select the virtual machines in the group of the highest correlation and allocate it on the target physical host. The experimental results and performance analysis show this strategy leads to a further improvement compared with the old migration strategies in CloudSim4.0. This strategy is valuable for other cloud providers to build a low energy consumption cloud data center.

Abstract：In the 3D recognition method based on 2D multi-perspective, multiple 2D projected images can be used to represent the feature information of 3D model. However, the features of projected images from different perspectives are different, and the learning efficiency of the neural network is also different. The convolutional neural network can map the features of images, and this method can be used to analyze this problem. The importance of projection features of different perspectives in the convolutional neural network was analyzed in the mix-view data set, and the collection density of the mix-view data set was optimized according to the different importance. The final experimental results show that the classification accuracy of 2D images generated from different perspectives is different, among which the classification accuracy of overhead projection is the worst, and the optimized data set achieves the optimal classification accuracy in the same neural networks model.

Abstract：In order to improve the generalization performance, most of the current mutual information feature selection algorithms have not optimized the specific application field, and electronic nose sensor array optimization is a special feature selection problem. It is difficult to search for the optimal feature subset using traditional algorithms. Combining the special redundancy and unique sensitivity of the gas sensor array, this paper proposes a mutual information feature selection optimization algorithm based on the performance of the electronic nose sensor. The influence of the sensor characteristics on the recognition accuracy is verified through two different electronic nose public data sets，which proved the effectiveness of the proposed algorithm.

Abstract：Domestic gas drilling construction process has begun to use microwave relay transmission mode to build monitoring while drilling system. However, the supporting LWD monitoring software applied in this construction condition is not mature, the data analysis time is long, the measurement transmission rate is low and there is no channel monitoring function, which restricts the development of gas drilling. According to the actual needs of gas field, this design adopts event driven software architecture to develop supporting MWD monitoring software, which can conduct human-computer interaction of various engineering parameters in the process of gas drilling in various ways such as curve, value, scale, etc., and can supervise microwave channel in wellbore, which provides software support for real-time monitoring while drilling and post drilling analysis and research of gas drilling. Moreover, the application of the software greatly improves the test transmission rate of the monitoring system, and the maximum measurement transmission rate reaches 1.194 kb/s. The introduction of channel monitoring function can increase the number of microwave repeaters entering the well and make the logging transmission depth exceed 3 000 m, which basically meets the needs of current gas drilling while drilling monitoring.

Abstract：Nuvoton′s Mini52(Mini51 series) microcontroller is used as the main control chip. By using the performance of the 6-channel PWM signal that the Mini51 series microcontroller can output, the sampling resistor is used to collect the terminal voltage to obtain the current of the DC bus. The zero point method(EMF) is used to control the motor to realize the dual closed-loop control of the BLDCM without position sensor. The serial communication is used to complete the servo system of the brushless DC motor, and the test is carried out. The test results show that the Mini51 series single-chip microcomputer can respond to high-speed serial communication in real time and the servo system has good speed controllability, which can meet the needs of modern industrial production real-time control.

Abstract：The switching point of the buck converter is the main object of external tuning. With the increase of the switching frequency and the increase of the load current, the parasitic inductance in the chip has an increasing influence on the switching voltage waveform, and the external buffer circuit is used as the adjustment switch. The waveform tool plays a key role, and only relying on experience to design the external buffer circuit can no longer meet the design requirements. By analyzing the working principle of the buffer circuit, combined with simulation results and engineering practice, this paper analyzes the voltage regulation performance and efficiency loss of the buffer circuit, and provides a quantitative analysis method for the design of the peripheral buffer circuit.

Abstract：Embedding non-intrusive load identification technology in the power supply entrance is conducive to promote building energy saving, realize power grid load forecasting, develop intelligent buildings and improve the construction of smart grid system. Therefore, this paper proposes a non-intrusive power load identification method based on directed acyclic graph support vector machines(DAG-SVMS). Firstly, the event detection of power system bus current signal is carried out. After the transient event is detected, the transient current waveform of the target load is separated and the features are extracted. Then, the features are input into the pre trained DAG-SVMS model for classification and identification. In order to improve the performance of the classifier, particle awarm optimization(PSO) algorithm is used to optimize the parameters of the DAG-SVMS model. In order to reduce the cumulative error, Gini index is proposed to optimize the node order of DAG-SVMS. The experimental results show that the proposed method has high recognition accuracy, fast recognition speed and feasibility.

Abstract：This paper introduces a WiFi digital RF front-end architecture which can be used for ASIC. It is compatible with 802.11b/a/g/n/ac/ax protocol in 2.4 GHz and 5 GHz channels. Combined with analog receiving circuit, ADC and demodulator, the functions of signal type differentiation, power locking of different types of signals, filtering and resampling of signals with different bandwidth are realized.

Abstract：This article proposes an AC strobe light based on a high power factor hybrid converter. This kind of high power factor hybrid converter can work in a wide voltage range of AC 80 V~250 V. The single-stage converter can achieve boost and step down, and the power factor is very high. The reverse voltage of the switch tube is the same as the output voltage. Therefore, the switch tube works at a lower reverse voltage, which is especially suitable for high-voltage output occasions to improve the stability and reliability of the circuit.The strobe lamp designed with a high power factor hybrid converter can realize the unitization of 110 V/220 V products, so the product is easy to maintain and manage, and improve management efficiency.