Signal and Information Processing, Networking and Computers

Due to its unique open source Android system has become a leader in the field of smart phones, allowing researchers to conduct a multi-angle study of the Android system. However, Android system has become malicious code attacks preferred target because of its open source features. For the existing detection scheme in terms of feature extraction due to the selection of too few types of features, the selected features contribute little to the classification accuracy of the classifier is not high and so on. This paper proposes a combination of dynamic and static multidimensional mixed feature extraction scheme, compared with the extraction scheme which only analyzes the authority and the function call, this paper extracts twelve types of features, which reflect the behavior of Android application from multiple perspectives and improve the comprehensiveness of feature extraction.

Merkle, a well-known cryptographers, is the first to give the conception of Merkle Tree authentication, which is that a complete binary tree is used for authentication and signature of a large amount of data. In recent years, Merkle Tree applied to wireless sensor network is a hot research subject, which has already been paid much more attentions by many researchers. To handle a large number of sensor nodes, Merkle Tree will cause several practical problems such as high computing complexity, the huge memory consumption, high load of communication, validation delay, etc. Grading Merkle tree effectively relieves some of these problems. But with the rising number of sensor nodes and broader geographical scope, wireless sensor networks based on grading Merkle Tree have emerge a lot of problems such as scalability, robustness, flexibility, security, etc. In this paper, wire sensor network struction based on hierarchical Merkle Tree is proposed to solve the above problems.

Owing to no need for prior knowledge of signal, blind spectrum sensing has received wide attention. Covariance Absolute Value (CAV) detection algorithm, one of the most popular blind sensing algorithms, considers the correlation of signal samples. However, its detection performance is restricted by the uncertain threshold calculation. To optimize the performance of CAV, we propose a new method based on a new statistic and goodness of fit test. The statistic is constructed from the off-diagonal of covariance matrix firstly, then Anderson-Darling (AD) test is used to estimate the existence or absence of primary user. The proposed method not only achieves blind detection but also improves the sensing performance of CAV. Experimental results manifest the effectiveness of the proposed scheme.

Time division duplex (TDD) massive multiple-input multiple-output (massive MIMO) has been regarded as a fantabulous technology for fifth generation wireless cellular networks (5G) to improve spectral efficiency (SE) and network performance using tens of hundreds of antennas and terminals attached to the base station (BS). On the other hand, the concept of pilot contamination (PC) is believed to be an acquainted gainsay in massive MIMO caused by channel estimation error, which disrupts and demarcates these required objectives. This paper projects an innovative strategy for pilot allocation i.e. cell splitting and sectorization based pilot assignment (CSS-PA) strategy to mitigate the PC. This strategy is based on exploiting signal to interference plus noise ratio (SINR), the users are first categorized into cell center and edge zones, followed by the subsequent sectorization of edge zones of each cell. Then, each cell center zone users are allotted with identical pilot sequences. On contrast, for edge zone sectors, the appointed sequences are mutually orthogonal in different cells. With the assistance of mitigated PC, we have ascertained the approximate system capacity, which shows precision for the unlimited number of antennas at the BS. The outcome of simulation reveals that our proposed CSS-PA strategy would effectively weaken PC. Moreover, the proposed idea has achieved higher system throughput, low Mean Square Error (MSE) and Normalized MSE (NMSE) at higher signal to noise ratio (SNR) and maximum number of BS antennas in comparison to the traditional pilot sequences allocation strategy with the marked sequences reuse rate of one or three.

A new cascade framework of real-time automated vehicle license plate recognition (VLPR) deep learning-based for intelligent transportation system applications in smart city is proposed. Our workflow includes two steps, which starts with localizing the license plate in the image captured from cameras by using you look only once (YOLO)9000 and then recognizes the whole license car plate characters leveraging convolutional neural networks (CNNs) without segmentation. We also investigate new networks that present the effect of the convolutional kernel size, depth, and width of CNNs on recognition performance. Even though these two steps are trained separately, we joint them together for testing images. The automated VLPR system that we proposed not only can display excellent performance even under the bad condition, but also be utilized in real-time deployed efficiently to GPU. In our experiments, 99.98% of recognition is achieved for whole Chinese car plate characters and the test-time for one image is only 17.25 ms.

License plate recognition system is widely used in real life, such as toll stations, parking lots, crossroads, etc. These specific applications can effectively alleviate traffic jams, save labor costs, improve efficiency, but it also plays an important part of the intelligent transportation system. At present, most of the license plate recognition systems use computer vision and image processing technology for license plate character segmentation, then character recognition. The research goal of this paper is to use deep learning algorithm combined with transfer learning to improve the generalization ability and accuracy of license plate recognition system than traditional methods. In particular, first, we use the Xception network to train license plate data with weights randomly initialized. Next, Transfer Xception model for image classification with weights trained on ImageNet to this task and train license plate data again. Finally, we compare the accuracy and efficiency of license plate recognition system between these two models and other deep leaning networks.

In this work, we propose a scheme using deep convolutional neural network (CNN) to detect and recognize vehicle license plates in complex natural scene. In particular, first, we propose to leverage the target detection method which named you only look once (YOLO) based on deep learning to detect the license plates. We optimize the network structure and train a 30-class CNN which can perform real time detection. Next, we combine the advantages of Dense Convolutional Network (DenseNet) and Residual Network (ResNet) and propose a simple, highly efficient network model named RDNet to recognize the license plates. Last, we concatenate two well-trained networks to detect and recognize license plate with high accuracy. The proposed scheme based on deep CNN needs free segmentation and the whole process needs no manual intervention. Extensive experiments verify the effectiveness and robustness of our proposed scheme, and the recognition accuracy achieves 99.34%.

This paper proposes a new method for segmenting Chinese license plates in which the license plate segmentation is performed using powerful deep learning techniques instead of traditional digital image processing techniques. Firstly, input license plate image is preprocessed using traditional digital image processing techniques: input image is converted into gray scale, and then skew detection and correction is performed. Secondly, license plate is segmented and characters are separated using a well-trained convolutional neural network (CNN) so that each character is in its own image. Those characters can be later recognized and classified using any character recognition module. The proposed license plate segmentation method is straight forward, less complex, and can be considered as a good alternative for some traditional digital image processing license plate segmentation methods. Also the main concept of proposed segmentation method has good extensibility so that it can be extended to any kind and format of license plates easily.

This paper focuses on the detection and recognition of Chinese car license plate in complex background. Inspired by the success of Deep Convolutional Neural Network (DCNN) and Recurrent Neural Network (RNN) in the field of object detection and image recognition, we propose to apply the YOLO detector for license plate detection, and Convolutional Recurrent Neural Network (CRNN) for recognition, which achieves state-of-the-art recognition accuracy. Firstly, we trained YOLOv2 and YOLOv3 for license plate detection, and compared their detection performance. Secondly, we designed and trained a network, named as CRNN-12, for license plate recognition. CRNN-12 contains a DCNN which is used to extract features and a 2-layer bidirectional Gated Recurrent Unit (GRU) which is used to decode the feature sequences. Connectionist Temporal Classification (CTC) loss function is used for the purpose of jointly training DCNN and RNN. The benefits of this approach are as follows: (1) It realizes end-to-end recognition without segmentation; (2) GRU can make better use of contextual information of license plate images, which leads to improved recognition accuracy; (3) License plate with different number of characters can be recognized by one network.

Vehicle-logo recognition, consisting of vehicle-logo location and its classification, is an important application of object detection in intelligent transportation. In this paper, we adopt the strategy of integrating Faster-RCNN model with two different convolutional neural networks (VGG-16 and ResNet-50) respectively and prepare a vehicle-logo images dataset containing 4000 vehicle images with different angles, backgrounds and resolutions for 8 different vehicle logos. In our experiments, the better mean Average Precision result of 94.33% is achieved in spite of the small proportion, huge intra-class variability and complex external environment of vehicle logos in the images, which shows that the methods based on Faster-RCNN can be used to recognize vehicle logos of road-monitoring vehicles and have good robustness. Integrating Faster-RCNN model with VGG-16 is better than ResNet-50 in the dataset we prepare, which illustrates the deeper network may not be the better for different recognition tasks with different amount of data.

Traditional telemetry (TM) process scheme makes the software high level coupled with the requirements, and hard to be reused. After analyzing and researching spacecraft TM requirements and spacecraft software component technology, we propose a general TM process scheme based on software components technology, dealing with TM parameter setting, data collection, data saving and download schemes for different space mission. The general TM process scheme can meet different TM requirements according to setting different parameters and assembling appropriate software components. After research, the result shows that the scheme has excellent Mission adaptability, can reduce the system develop-time-consuming and cost.

In this paper, the development and technical features of Micro-small Satellite are introduced. According to the functional requirements and restricted conditions, the Attitude and Orbit Control System (AOCS) is designed for one occultation sounding satellite constellation. In the design of AOCS, the Integrated Electronic System (IES) based on the integrated electronic technology is realized, which consists of many hardware function modules and software function modules, and some miniaturized products are adopted such as micro star tracker and Micro-Electro-Mechanical System (MEMS) Gyro. By the dynamics features of the occultation sounding satellite constellation, a dynamics simulation system is designed to test the AOCS at the same time. According to the result of simulation, it can be known that the AOCS based on the integrated electronic technology satisfies preferably the strict requirements of the occultation sounding satellite constellation.

In order to carry out the on-board software development more efficiently, a virtual test platform of OBDH software based on QEMU emulator is designed and implemented. Based on dynamic binary translation, the virtual on-board processor and peripheral device are simulated to run OBDH software in development platform, and history machine instructions are recorded for on-board software debugging. Combined with the chip driver and terminal application, the system data flow is simulated to provide the input data of on-board software, and the system log is record in order to provide a single monitoring window by stratified processing and real time comparison of the output data. The virtual platform has been used in some OBDH software development, the result shows that it can provide the running environment of on-board software, provide more software debugging means, simulate and record the system data flow, so that on-board software test can be done before the on-board hardware is ready. The virtual test platform can improve the efficiency of on-board software development.

In order to meet the requirement of high speed and real time data transmission of large field of view space remote sensing camera, a reliability transmission strategy on the base of TLK2711 is proposed. First, the communication mechanism of TLK2711 and the 8B/10B coding principle are analyzed. According to the minimum transmission protocol, the paper develops a simplest protocol used to high-speed serial transmission. Then, in order to further enhances the signal integrity in the transmission process, a half-ECC algorithm is applied and it is easy to implement in hardware. Finally, the system experiment is performed on a high-resolution multispectral TDICCD camera. Experiments verifies that the proposed transmission strategy can effectively reduce the number of transmission lines to 1/24 of the traditional transmission lines, simplifying the camera interface, and also complete point-to-point serial transmission with a data rate of 1.76 Gbps. It can be seen that the single particle flip event can be effectively corrected during the data transmission process through sending self-calibration graph. The proposed transmission strategy ensures the signal integrity of remote sensing camera image transmission, and meet the miniaturization and light-duty of the system.

Global navigation satellite system (GNSS) is of significance for orbit determination of Low Earth Orbit Satellites. Due to relatively low accuracy of code observations and high kinematic challenges, the accuracy of orbit determination by using GNSS will be limited. With the combination of the emerging constellations BDS and the modernization of GPS, the number of satellites available for users is increased. In this contribution, a modified reduced-dynamic EKF-based GPS+BDS combined model is proposed to improve accuracy of orbit determination. The proposed method uses dynamic model of orbit and EKF algorithm to smooth the GNSS positioning results. Meanwhile, this method also uses the GPS+BDS combined model to further improve the GNSS positioning accuracy. The proposed method is tested by simulated date by STK software. The simulated results show that positioning accuracy of the proposed algorithm improves almost by 45% compared with traditional algorithm. Additionally, the proposed algorithm has more robustness than the traditional algorithm.

This paper studies the application of the satellite interference multispectral imaging technology in building safety monitoring. The satellite interference multispectral imaging monitoring technology is to measure the radar interference by the observed values of amplitude and phase by repeated orbits or double antenna systems. Compared with the traditional measurement methods, the interferometric SAR satellite imaging technology completely overcomes the defects of low data acquisition ability and poor stability of the system, and has the characteristics of high real-time and large detection range. The use of satellite interference multispectral imaging technology to monitor the safety of buildings is mainly to monitor the size of the building’s settlement and the type of settlement. Combined with the relevant building structural safety regulations, the impact of building settlement on the safety of buildings is inferred. Due to the mature development of D-InSAR technology, the use of ERS data for ground subsidence monitoring can reach an accuracy of 10 mm, which opens up new ideas for building deformation subsidence monitoring.

The platinum resistance temperature detector, PRTD, is the best choice for high accuracy temperature measurement system in spacecraft. Interfered by factors such as constant current source drift and thermal electromotive force of lead wires, the traditional approach used for precision temperature measurement utilizes a constant current loop in a four wire Kelvin connection, which is hard to achieve a measurement accuracy of 0.05 °C. A high precision temperature measurement method for spacecraft based on reference resistor is proposed in this paper. The voltage difference between the PRTD and the reference resistor with the same current flowing through the PRTD is measured. The bias voltage is eliminated by designing resistance of reference resistor, which maximizes the use of AD converters. To minimize the effect of excitation current variations, the standard resistance channel in parallel with PRTD is used to measure the excitation current in real time. Reference resistors and standard resistances with low-temperature coefficients (<5 PPM) are installed in the temperature control objects to eliminate system temperature drift. According to temperature measurement requirement with the accuracy of 0.01 °C for a temperature range of +10 to +30 °C, two different design scheme is proposed. The innovative measurement method proposed in the paper can realize a temperature measurement uncertainty of 0.5 mK. The innovative method proposed in the paper can be used for high resolution, narrow range, accurate temperature measurement of spacecraft.

Installation error of optical communication terminal is introduced during the laser communications terminal installation process, and it makes the establishment of satellite-ground communication link more difficult, due to the laser only allowing for dozens of micro arcs coverage angle. In this paper, a method using in-orbit test data for laser pointing error reducing is proposed. Based on the terminal coordinate system, the cubic mirror transition coordinate system and the satellite body coordinate system, the transition matrix of the satellite body coordinate system and the terminal reference coordinate system based on the coordinate transformation is developed. Through in-orbit laser communication test, errors are obtained which is between the terminal coordinate system and the cubic mirror transition coordinate system, and between the cubic mirror transition coordinate system and the satellite body coordinate system. Finally, the terminal installation error is rectified. The proposed method is applied in in-orbit communication experiment, and shows it effectively reducing the laser antenna pointing deviation.

The double axis Digital Sun Sensor (DSS), product of Beijing Institute of Control Engineering (BICE), is widely used in spacecraft in China. The open flag current of DSS is one of telemetry to provide the information about sunlight intensity. Every axis has telemetry of open flag current called ‘X axis open flag current’ or ‘Z axis open flag current’. Based on the in orbit data of project SJ-6 which launched in 2008, the damping tendency of the open flag current of DSS is analyzed in order to find how the performance of photoelectric cell unit will be changed by sun radiation in long term on board working. According to the analysis, during 6 years worked in orbit, X axis has attenuated about 13.7%, and Z axis 16.0%.

The load chassis is the main component of the installation of fly-around satellite and repressurization system of spacecraft. Its stiffness and strength take the important role for the normal work of the fly-around satellite and repressurization system of spacecraft. The initial design is not good in the mechanical performance test of the chassis. Based on the existing structure, several optimization designs for the dynamic performance are proposed, The final design scheme is determined by comparing the results of finite element analysis and operability, which was verified through the product test and the whole spacecraft vibration test. The fundamental frequency of the structure is increased by nearly 8 Hz, Although the response value does not decrease largely. The successful launch of the spacecraft and the satisfactory shooting of the fly-around satellite indicate that the optimization design of the loading chassis is correct and reasonable. The optimization design method also provides experience for other spacecraft structure design.

The digital signal processor (DSP) is mainly found in the electronic systems of spacecraft, which is very important as a controller and calculator in satellite optical communication. The space radiation environment can cause the function degradation, parameter drift and even device failure of DSP devices, so we studied the effect of total radiation dose on DSP. We establish the model of parasitic body tube by the radiation theory and the theory of superposition of the unit parasitic transistor to obtain the total dose effect model of the n-channel MOSFET and simulate the total radiation dose effect on typical commercial type of the DSP, the total failure dose node of this commercial DSP is obtained using this method. Further on, the method established in this paper is also applicable to obtain the total radiation dose effect on other type DSP. The total radiation dose effect data of DSP is obtained by simulation can be greatly reduced for spacecraft design.

Due to the outstanding capability of control torque amplification and smoothly continuous output, Control Moment Gyro (CMG), as a new type of high-performance attitude control actuator of satellite, therefore has been adopted by more and more high-performance remote sensing satellites to implement the high temporal and spatial resolution of in-orbit application requirements. Based on the mission requirements of a certain spacecraft, this paper gives a detailed analysis of the utilization capacity of satellite actuators’ angular momentum envelope under different CMG-group configurations. Within the design constraint of determined actuator configuration selection, the high-reliable ability of in-orbit singularity avoidance and high-efficient utilization ability on actuators’ angular momentum are realized by designing and applying singularity avoidance method based on vector control, which lays a foundation for high-performance maneuvering of the satellite platform. In addition, in order to enhance the application effectiveness of satellite payloads, the proposed adjustment strategy of in-orbit bias angular momentum is applied. Moreover, the significant improvement on the maneuverability of the whole satellites specific direction is realized. Furthermore, the requirement that the single array camera can obtain the stereoscopic images of a single target with multiple shots in the same orbit is satisfied.

Low frequency is becoming more and more important for the operators to deploy LTE network because of its wide coverage. And before deploying, it is very essential to analyze the user equipment (UE) amount that camped in the operator’s LTE network. LTE FDD band 8 (L900) terminal in China Unicom’s LTE network was instituted in this paper. UE’s L900 software capability analyzation method was proposed by exacting the signaling messages from S1 interface. Then the L900 software capabilities of the UEs were used to form the database, which could be utilized to analyze all the UEs that support L900 statistically. Statistical results showed that, Apple supported L900 for China Unicom best. And all of the apple 4G terminals supported L900 in hardware, and about 80% supported L900 in software. It was shown that about 62% and 48% 4G terminals of the collected subscribers supported L900 in software and hardware respectively. Therefor, it is very essential for the operator to require the terminal manufacture to support the LTE frequency at least one year earlier before deploying.

This paper discusses how to utilize the big data technology to discover some key problems concerning the data value during the construction of information management system of Center for Disease Control and Prevention (CDC) Laboratory in Guizhou Province. Depending on the laboratory information management system (LIMS), all levels of CDC in Guizhou province jointly carry out related business work within the framework of networked laboratories to realize the standardization of laboratory data and the comprehensive utilization of data across regions and levels. The realization of “unified deployment, unified construction, unified technology, unified management, unified maintenance” of limited resources integration, effectively avoids duplication of construction.

In this paper, a new algorithm based on telecommunication data for the mapping of regional business potential is introduced. This paper first describes the characteristics of telecommunication data and the data that needs to be integrated and prepared before geographical portrait analysis. Second algorithm design model is given, and the model includes three steps: the first step is to predict user resides the purpose of, the first step to set 7 class portrait geographical features, and then according to the POI data using the Bayesian model and the maximum a posteriori probability estimation method to get a specific geographical area’s portrait of functional characteristics, and to compare the user reside features to get the user reside purposes. The second step is to predict the potential demand of users in the geographical area. The third step is to analyze the potential demand of geographical regions. In this step, the normalized difference between the potential demand of users and the functional characteristics of geographical regions is analyzed to obtain the potential demand portrait of geographical regions. Based on the data extracted from the real environment, this paper analyzes the results, and gives the predicted value of the user’s residence purpose, the potential demand value of the user and the potential demand index of the geographical area. This paper verifies the feasibility of the method, by which the feature description and the commercial potential of geographical region portrait can be predicted.

The prediction of User Equipment replacing is worth of research both for telecom operators and mobile phone companies. This paper designs a machine learning prediction of User Equipment replacing (MLPUser EquipmentC) architecture and a data mining algorithm called CQSFL-LR (Composite-parameter Quantum-inspired Shuffled Frog Leaping Logistic Regression), aiming at researching the factors and their weight respectively of a telecom user whether will replace his cellphone or not. Experiment shows the proposed CQSFL-LR algorithm has better performance in accuracy and precision compared with traditional Logistic Regression, proving the superiority of CQSFL-LR. The experiment also shows MLPUser EquipmentC architecture can predict User Equipment replacing, providing marketing guidance to telecom operators and mobile phone companies.

When handling large-scale national road networks, existing operation monitoring systems suffer from many limitations, such as inadequate monitoring facilities, unbalanced regional distribution, low quality of monitoring devices, abominable environmental conditions and incapability of the maintenance staff. This paper uses massive mobile signaling as a new information source to facilitate the monitoring of road network operations. Mathematical modeling is employed to analyze the business requirement for the monitoring of road network operations, and investigate the spatial distribution and movement patterns of mobile phone terminals in the mobile communication network. Combining those information and those information that obtained from traditional information sources offered by the transportation management organizations, the proposed method can monitor the status of high-speed traffic in real-time through multi-source data fusion analysis and visualize it on a GIS map.

Information technology and economic society are deeply integrated, which promotes information systems extending from single application and single organization to cooperative management and services crossing level, region, system, department and business, and then causes explosive expanding of data. As data are collected from multiple sources, they are heterogeneous and low-quality, and they are blocking information exchange and interoperation. Data quality problem has become an important factor which seriously hinders the improvement of data analysis and decision support ability. To solve the data quality problem, the most important thing is to manage and measure the quality of data. Base on the existing researches such as quality management and data quality, this paper proposes a data quality management process framework (DQMPF) and a data quality problem and measurement model (DQPMM). Furthermore, taking the international trade document as an example, this paper applies the proposed innovative theories to reveal the document data quality problem.

User perception is the key for telecom operators. In order to keep good user perception, telecom operators employ network optimization and construction. In this paper, a novel user perception aware network management (UPNM) algorithm is designed for LTE/LTE-Advanced cellular networks. Initially, UPNM algorithm integrates a series of telecom data, which relates to users and cellular networks. On the basis of data, user perception is analyzed to find the cells with low quality of service (QoS). Then, UPNM algorithm analyzes the network interference and coverage and takes methods to deal with the interference and coverage problem. Finally, both the cell total service and capacity are analyzed, in order to manage the network capacity. We implement the proposed UPNM algorithm to the realistic cellular systems in a Chinese city. The UPNM algorithm can help telecom operators manage the mobile cellular networks.

As we know, the mobile network cells along the high-speed railways serves both the subscribers on the train and the subscribers on the ground, but the two kind of subscribers move differently, therefore the traditional KPI-based analysis method cannot truly reflect the perception of high-speed railway subscribers; However, the on-board DT testing has poor timeliness and limited types of terminals. It is necessary to find a way which is suitable for reflecting the true perception of subscribers on the high-speed rail in real time. Based on the analysis of big data, this paper first proposes a way to evaluate the true perception of subscribers on the high-speed rail. First of all, we distinguish the subscribers on the train from the subscribers on the ground, then evaluate the true perception of subscribers on the high-speed rail by their Signaling plane and user plane data.

The measurement report (MR) contains key information such as the received signal strength of the user terminal in the network, which can reflect the coverage and interference of the LTE network. With the advantages of low cost, large amount of data and high reliability, measurement report has been widely used in practical network optimization. In this paper, a centralized distribution area of weak coverage sampling points is obtained based on the MR and combined location information from the actual users in network. Through the continuous iterative adjustment of antenna parameters, the optimal setting of antenna horizontal direction and vertical down-tiltangle are found, which can optimize the coverage performance of the entire cell. In order to verify the performance of the antenna optimization algorithm in this paper, at the end of the paper, the application effect of the scheme in the practical network is given. Through the verification, the optimization scheme proposed in this paper can effectively improve the coverage performance of the weak coverage cell.

An ultra-dense network scenario is a scene which a large numbers of people assemble in limited areas to generate centralized broadband data traffic requirements. Ultra-dense networks generate enormous traffic pressure, and traditional network capabilities are far from enough to accommodate the user’s needs. Based on the description of ultra-dense network structure, this paper analyzes millimeter wave radio spectrum, high gain beam forming, physical layer frame structure, resource concentration and Mobile Edge Calculation techniques. The cooperative technologies required by overlay and interference symbiosis in the dense network architecture, as well as the access control technology of centralized access, are analyzed and discussed comprehensively.

With the formal commercial application of VoLTE, the number of 4G users has increased rapidly, which has brought great challenges to operators and brought new challenges to LTE wireless networks. The commercialization and popularization of VoLTE brings the demand for more novel and more diversified multimedia services. The traditional wireless network analysis method has been unable to meet the needs of different features and different scenarios. In view of the increasingly personalized users’ needs for perception and experience, the wireless network analysis method should be adjusted according to local conditions, so as to realize the optimization of network structure. Therefore, a new LTE value evaluation system is proposed to identify high value residential areas and optimize the network structure. The system uses four dimensions of consumption capacity, call volume, network request and displacement, subdivides the user value, and divides the geographic grid value according to the user location information. At the same time, it analyzes the network bottleneck for different value users and grid, and gives the optimization and planning scheme.

This article describes the Double Deck flexible bandwidth networking solution. By using the strict filtering and RB compression technologies and the innovative LTE cell orthogonal algorithm, two LTE standard bandwidth cells are overlaid to make full use of scattered spectrum resources. This feature effectively improves the spectral efficiency of the network, enables GSM and LTE to be properly used on the 1830 MHz–1860 MHz frequency band, expand the LTE bandwidth while ensuring continuous coverage of the GSM1800 in the S111 configuration. In addition to high-quality voice and data services on the existing 2G/3G network, the frequency reuse and spectrum utilization efficiency are improved to further improve the coverage and capacity of the 4G network. In this article, analyzes the KPIs of the 125 trail sites in Hongkou District of Shanghai, to evaluate the impact of the solution on the live network, and introduce the implementation solution and the KPI trend are described in detail.

With the continuous development of wireless network, network management and maintenance are faced with a lot of challenges like massive network elements, heterogeneous systems, and multi-vendors. In response to this situation, the industry proposed the concept of SON (Self-Organization Network), which is designed to make the planning, configuration, management, optimization and healing of mobile radio access networks simpler and faster. ANR (Automatic Neighbor Relation) is one of the important functions of SON and also a key technology for the optimization of LTE neighbor cells, but ANR has limitations in its scope of implementation. The article proposes a new method for LTE automatic neighbor cells optimization based on MR (Measure Report), transmit by user equipment, which is more simple and efficient. Our proposed method can accurately identify issues such as neighbor cells missing, PCI (Physical Cell Identifier) confusion, ultra-distant neighbor cells, and redundant neighbor cells. The results indicate that our method improve the neighbor cells optimization project efficiently.

With the development of mobile communication network, the density of networking is getting higher and higher, and there are multiple network coexistence of 2G, 3G, 4G, NB-IoT and so on. The interference problem between networks is more prominent, which seriously reduces the network performance and user perception. Traditional jamming recognition mainly depends on expert experience and artificial judgement. Manual processing cycle is long and efficiency is low. With the increasing number of 5G ultra dense networks, small base stations and even home base stations in the future, how to identify communication network interference quickly and effectively has become an urgent problem to be solved. Therefore, the advantage of deep convolution neural network (deep convolution neural network) to deal with the interference of large data in the communication network is studied, and it is applied to the domain of interference image recognition, and a method based on depth learning for interference image recognition of communication networks is proposed. Based on interference image simplification and jamming image recognition algorithm, a depth model based on interference spectrum recognition is established. The combined learning and overall optimization of the deep convolution neural network greatly improves the efficiency of interference recognition, thus optimizing the quality of the network and improving the user’s perception.