Abstract：For heterogeneous wireless environment, this paper proposed a network selection algorithm based on load balancing. The strategy “terminal auxiliary, network decision” is adopted. Firstly, the network reference vector is got by using the analytic hierarchy process(AHP) modified weight. Then, terminals select the set of access networks according to the geometrical relationship of vector. Finally, the network side allocates the suitable access network, which is from the set of access networks, for the terminal based load balancing. The simulation confirms the algorithm is validity.

Abstract：With the development of integrated modular avionics, the part of RF processing and the part of digital signal processing in the sensor system are designed as general modules. As the node of data exchange between the general RF processing modules and the general digital signal processing modules, the data exchange network system implements the function of configuration and reconstruction of communication channel resources in avionic devices. Based on the LVDS technique, the realization and the working principle of the data exchange network system are introduced.

Abstract：Considering the serious routing-flooding and poor self-healing performance of the existing wireless mesh network routing algorithms, in this paper, a centralized self-healing routing protocol(CSRP) combined with centralized routing and multi-path strategy is proposed. In this algorithm, the root node masters the newest network weighted topology through the uploaded link state information from each node, and provides the best-metric route for source-destination pairs. When a route breaks it can recover seamlessly thanks to the multi-path backup strategy. In addition, root node switching mechanism based on the link quality is joined to ensure the reliability of the routing information’s application and distribution. Simulation results show that CSRP suppresses the flooding, and improves the performance of the network effectively. CSRP can save routing overhead at least 67%. Meanwhile, with CSRP, the average delay can be reduced by 45% and the average packet delivery fraction can be boosted by 8%, compared to the existing classical algorithms.

Abstract：It is more difficult to assess the effectiveness in communication networks, thus this paper proposes an ADC model based method to assess the effectiveness of communication networks. In the ADC model, row vector, matrix characteristics and reliable communication support capabilities can be selected as the key indicators. The characteristics of the communication networks can be fully reflected by these indicators to carry out a scientific assessment. The theoretical analysis and experimental tests prove that the proposed method can effectively evaluate complex network performance metrics, and give managers a reasonable basis for decision making.

Abstract：Micro/Nano-Electro-Mechanical(MEMS/NEMS) switch refers to the mechanical switch which its feature size in the micrometer or nanometer scale, it is an emerging technology, has the advantages of small size, fast speed and zero leakage,it is becoming a hot topic in recent years.The applications of MEMS/NEMS switch in RF devices and logic circuits are introduced firstly, and summarize the classifications and characteristics of MEMS/NEMS switch. At the same time,the research status of MEMS/NEMS is summarized systematically,including some breakthrough in the size, threshold voltage, stability, material,etc.Finally,the development trend of MEMS/NEMS switch in the future is discussed,the problems and obstacles are described, such as the influence of adhesion under nanoscale, the realization of the existing technology,etc.

Abstract：This paper proposes a new design,that is the way in which mobile vehicle-mounted terminal automatically adapt to wireless communication module in Linux. In order that module can be intelligent recognized，firstly，the wireless communication module gets power by controlling hardware commutator board and time sequence. Then it is tried to read the PID/VID from the wireless communication module，and the version type of it is gotten by AT command. Finally，the module is initialized through the polymorphism function of the object-oriented language. This design is achieved by keeping the vehicle mobile terminal and code，while，the variety of different function of the wireless communication module still can be used.

Abstract： OMAP3530 is a high-performance embedded microprocessor designed by TI company for open multimedia application platform. A new type communication method is designed between the ARM and DSP based on DSP/BIOSTM LINK software framework. Particularly, Linux operating system is embedded on ARM side, to implement decoding, human-computer interaction and data management tasks. DSP/BIOS real-time operating system is embedded on DSP side, to execute rough interpolation, cutter compensation and logic control tasks. The system can realize quasi parallel operation of multi task through task interrupt service and time slice round-robin method and complete communication between ARM and DSP through dual-core communication. The system uses master-slave computer control mode, which greatly enhances the real-time performance and the interpolation of CNC system.

Abstract： In this paper, a method for simulation and hardware implementation of phase noise channel of Ka-band communication satellite is investigated by adopting multiple-level IIR filter in IF to realize power-law distributed noise and obtain simulation of phase noise. The design method of IIR filter and simulation precision of phase noise are also discussed. Finally，a phase noise simulator is realized based on FPGA+DSP hardware and measurement result shows that its simulation range is superior to that of Aeroflex channel simulation.

Abstract： Forcusing on the requirement of airborne gigabit switch system configuration and Build-In-Test(BIT), a kind of embedded gigabit switch system is designed based on micro control unit(MCU). Focusing on requirement of software portability and scalability, object-oriented design based on unified modeling language(UML) is adopted, static model of software is established，and software flow char and design are given. Via design and implementation on the MCU platform, the hardware function is normal，and the software can well implement the configuration and BIT for switch system.

Abstract： For the application of deep-sea equipment in marine scientific research and in the field of resources exploration, a kind of landing control technology used for deep-sea equipment was developed and implemented, and a principle prototype was also completed. The principle prototype takes STM32 processor as the core and combines with altimeter, inclinator and step motor. Then with the adoption of innovative mechanism of "secondary leveling", the principle prototype realized the complex terrain of smooth landing and the leveling function in the environment of the laboratory pool. The test shows that it can accomplish those functions in a reliable way and meet the requirements of the design.

Abstract： To overcome the problem of less reliability and construction difficulty in centralized control system of stress monitoring, an approach of distributed control system based on wireless sensor network for stress monitoring is proposed. The system framework, hardware circuit and software design are described in detail. Based on distributed wireless ZigBee, terminal sensor node is designed to obtain the dynamic stress data in real time. The data in coordinator are uploaded to master computer through CAN bus and then to be processed. Testing result shows that the system has good real time ability, high reliability and can meet the practical demand of stress monitoring.

Abstract： For the problem in wave energy power generation process of getting the wave characteristic parameters such as frequency and amplitude in real time, STM32F4 series high-performance microcontroller, MEMS motion sensors and GPRS wireless transmission technology are used to design a real-time remote wave characteristic parameters monitoring system. This paper mainly describes the hardware structure of the monitoring terminal, the MEMS motion data integration algorithm, and the short-time Fourier transform analysis algorithm of wave′s characteristic parameters extraction. Practical application verifies the validity of the method.

Abstract： With the reduction of process dimension and the development of working frequency, the packaging technology of PCB and bonding wire are likely to impact on the electrical characteristics of the circuit, it is necessary to analyze signal integrity of the packaging technology. This paper introduces a surface-mount package TO-252 and model it by using HFSS. There are some factors that have an impact on signal integrity of the package,such as the material and thickness of PCB, the length of bonding wire, the arch height of bonding wire, the distance between two adjacent bonding wire, the number of bonding wires. They play guiding roles in the physical package.

Abstract： In order to improve the efficiency of implementing multi-task cryptographic algorithms, the characteristic of multi-core processing and the basic law about the multi-core connective structure——the law of Amdahl is analyzed in this paper. Then the structure model of multi-core processing is proposed, and the simulation and analysis of the parameters on affecting the performance of multi-core processing is implemented. The improced connective structure of 2D-Mesh and the implementation are provided based on the result. Finally the work efficiency of the improved scheme is verified by the simulation tool of VCS. The result shows that the plan can effectively accelerate the speed of message passing in multi-core system compared to the conventional structure of 2D-Mesh, proving the rationality of the improvement programs.

Abstract：As visual or command data is sent to LCD module via FPC or PCB, some problems like crosstalk, transmission loss and reflection may result in transmission errors, and the relevant simulating method is described. There are also some discussions about simulation of source lines and gate lines. Lastly, a set of testing circuit is designed to check the LCD module’s inrush current, which could be induced by power-on slew rate of power sources, or by power-on time interval between different power sources.

Abstract：The main contents of this thesis is to research into the method of which designing a linear power amplifier ,which is fully integrated, applied to the 1.95 GHz WCDMA transmitter by adopting the 0.18 μm SOI CMOS process of IBM’s research. The circuit structure consists of three class AB amplifier which are cascaded together. According to the simulation results, the CMOS RF power amplifier is unconditionally stable when the power supply voltage is 2.5 volts and the 1 dB compression output power is about 30 dBm with that the gain is about 28 dB. It also shows that the maximum power added efficiency(PAE) is about 42%.

Abstract： In order to avoid the damages to the body that owing to the incorrect postures and excess exercises in the process of movement and race, the article proposes the design of a movement monitoring system based on the human body sensor and Android technology to monitor the real-time parameters and postures of the human body in the process of movement. Using the Android platform to calculate calories consume and evaluate modeling of the movement posture in order to avoid sports injury and physical overdraft. Adoption MSP430 microprocessor as controller of the monitoring system, three axis acceleration of the human body can be measured by using MPU6050 three-axis acceleration sensor.Moreover, motion parameters of acceleration, speed, mileage and consumption of calories in the movement can be detected. The trajectory of the human body can be tracked and positioned.Through the identification algorithm of human body posture, it can evaluate human body motion and prejudge dangerous action. Experimental results show, system can realize real-time motion monitoring and tracking, protect of health of exercisers.

Abstract： By analysing the comfort situation of human arm in somatosensory control , based on fatigue theory, we put forward the somatosensory manipulator comfort control algorithms. Modeling for arm comfort zone , controlling robotic arm movement in somatosensory way, through the radial testing laboratories and dot point experiment for the comfort of the control arm movement parameter setting, we fit control the mapping function,and raise comfort control algorithms. By grasping motion tracking and remote objects experiment, the relative error of the measured comfort control algorithms in less than 5%, the absolute error of tip position(the end of mechanical arm) is less than 2 mm, improved the comfort of approximately 51.8% compared with conventional control algorithms , ensured the comfort of the user at the same time meeting the accuracy ,and improved the efficiency of somatosensory control.

Abstract： Based on the analysis of the structure characteristics and working principle of grinding active measurement instrument, against the problem of less channel of grinding active measurement instrument and urgently needing one machine with many measuring device, a multi-channel extension method of grinding active measurement instrument based on CAN bus is proposed, the multi-channel extension system solution is built, and the system software and hardware design is implemented. Experiment and analysis based on the system built realize the data real-time communication and interaction between grinding active measurement instrument and the measurement channels, verify the feasibility and validity of the extension scheme. Proposed extension method has high reliability, good expansibility, convenient operation, stable performance and low cost.

Abstract：In order to realize fusion processing of multi-source measurement data, system error registration was studied. Aiming at the complex electromagnetic environment interference, a real-time median estimating algorithm of slowly varying sequence was designed under the guidance of robust statistics theory based on the sorting of bidirectional linked list. On the basis of which a robust system error registration method based on fusion detection and real-time median estimation of the difference between multi-source measurement data was put forward. The unrobustness of traditional registration method such as the least square method was verified by test with measurement data. The test also indicated that the registration method proposed in this paper was effective and well robust both in time invariant and time-varying system error processing, avoiding large deviation.

Abstract：Due to such complicated layout of workshop as multipath effect led by obstacles and lack of reference tags distributed on the boundaries, location identification based on dynamic active RFID calibration(LANDMARC) algorithm, used in workshop locating, is not capable of selecting the nearest neighbors of the given reference tags effectively. To solve the problem, proposing an original workshop layout-oriented adaptive LANDMARC algorithm. The algorithm firstly introduces logarithmic distance path loss model to process the received signal strength indication(RSSI) from RFID readers, then filters the ones with feasible values out of all reference tags by method of comparison and inserts into the boundary virtual tags obtained by method of interpolation. Compared to traditional LANDMARC, the algorithm reduces the mean error from 0.39m to 0.16m, which clearly shows the effectiveness.

Abstract： In this paper, the vector network analyzer and X-band antenna are combined to be put on a track to synthesize aperture of azimuth direction, which builds a low-cost laboratorial ground based synthetic aperture radar system which also has the capacity of 2D imaging. Based on this system, the experiment of interference imaging of ground targets is conducted, and then the phase difference of two images is extracted, lastly, the deformation or displacement of targets can be computed based on the geometric modeling between phase difference and deformation quantity. Through the practical experiment, the ground based differential interferometry micro-deformation detection experiment system is demonstrated to be valid to measure millimeter-scale deformation.

Abstract： The existing graph mining algorithms in a cloud environment is difficult to carry out mining the high frequent patterns of a massive graph .To solve this problem, this paper has made the improvement to the SpiderMine algorithm, an improved SpiderMine algorithm is proposed based on the cloud(c-SpiderMine). Firstly, one big graph data into several sub graphs by minimum cut algorithm to minimize partition/merge costs. And then exploits SpiderMine to mine the patterns, which generating large patterns with much lower combinational complexity. Finally, a pattern key (PK) function is proposed to preserve the patterns, which guarantees that all patterns can be successfully recovered and merged. We conduct the experiments with three real data sets, and the experimental results demonstrate that c-SpiderMine can efficiently mine top-k large patterns in the cloud, and performs well in memory usage and execution time with different data sizes and minimum supports than the SpiderMine.

Abstract： It is found out that approximately one third of the chosen adjacent tags actually appear in triangles in the LANDMARC localization algorithm. Accordingly, the form of reference tags can be changed from rectangle to triangles in LANDMARC algorithm. The simulation comparison shows that triangle form of reference tags will improve the localization accuracy. It is more important that triangle form of reference tags will requires less tags, reduce cost, decrease signal interference, and enhance the stability of localization system.

Abstract：We propose a residuals statistics-based additive outlier detection algorithm for one-dimensional time series, The basic idea is using time series AR model for forward and backward fitting. In order to reduce the influence of outlier, we use data’s relative change rate to preliminary judge the outlier. According to hypothesis testing theory and Gauss distribution statistic testing, we find out the outliers. The simulation results show that the this method has good performance on outlier detection.

Abstract： For wireless communications system, cooperative relaying technique can improve system’s symbol-error-rate(SER) performance efficiently. This paper analyzed cooperative communication system with single DF relay and a cosed-form SER formulation and corresponding upper bound was derived by making use of a simple linear topology .The optimum power allocation and relay location were derived, which was aimed at minimizing SER of the DF system. Finally, the optimal joint adaptation of the parameter triplet was given by numerical optimization techniques. Theoretical analysis and simulation results indicate that, by using the optimal relay location and the optimum power ratio, the average SER for DF cooperative communication system can be optimized.

Abstract： Spectrum cartography is an emerging technique that can be used to discover spectrum holes in space. Distributed spectrum map can better adapt to local changes compared with centralized in cognitive radio networks. The secondary users generate the triangular network based on Delaunay triangulation growth method. Then spectrum situation map is generated by the local interpolation. Based on distributed spectrum map, this paper proposes a distributed hybrid access scheme for secondary users. Secondary users choose overlay or underlay access scheme according to the situation of spectrum information. The optimal power allocation scheme was deduced under the criterion of maximizing the system average capacity. Simulation results show a significant improvement in the system average capacity, in comparison to traditional detect-and-avoid algorithms.

Abstract：This paper investigates the performance of multiple-input and multiple-output(MIMO) antenna arrays systems including MIMO space-time correlation and antenna array configurations. In this letter, a closed-form expression for the spatial fading correlation function of a uniform rectangular array(URA) under various angular energy distributions in a three-dimensional(3D) multipath channel is derived. This article uses multiple signal classification(MUSIC) algorithm to estimate the spatial spectrum of the MIMO system direction of arrival signal. In this letter, the spatial spectrum generic formula of multiple antenna arrays is derived. According to the program computer simulations are carried out to verify the analytical results. Simulation results show azimuth spread(AS)is the main determinant of the spatial fading correlation between antennas, while at low azimuth spread values the impact of the elevation spread(ES) is mainly noticeable. The results show that when estimated the spatial spectrum of MIMO systems in the same parameters case, using three-dimensional uniform rectangular array is better than the other antenna array.

Abstract： Based on the BA scale-free network and combing with the characteristics of microblog user network, this paper constructs a network model for microblog rumor spreading. A novel susceptible-infected-removed(SIR) model is proposed, and by this model and numerical simulations, we investigate the dynamical behaviors of microblog rumor spreading. Research shows that, when the microblog user scale and the number of new connections between nodes increase, the final size of microblog rumor spreading obviously increase, but the critical threshold is greatly reduced. Also it is found the proportion of nodes with greater attraction degree and the probability of mutual following among microblog users do not influence the propagation characteristics of microblog rumor.

Abstract：There are problems such as time delay, packet loss in network systems. When it is introduced in traditional control systems as signal transmission medium, the control performance will be deduced. In this paper, a predictive control algorithm is used to study time delay in networked control system. Matlab simulation results show that the network predictive control can ensure good control output. This paper focuses on mathematical realization of predictive control method for in detail and it plays an active role in guiding the engineering applications.

Abstract： This paper analyzes the immunological characteristics of random immunization, acquaintance immunization and targeted immunization on a microblog user relationship evolution network. Based on this, according to the specific characteristics of microblog user relationship evolution network, a novel targeted immunization is proposed. The results show that when the node attraction degree obeys uniform distribution, the immune effect of random immunization and acquaintance immunization is basically the same. When the node attraction degree follows exponential distribution, the performance of acquaintance immunization is better. We also find that when the node attraction degree follows uniform distribution and exponential distribution, the targeted immunization obviously performs better than the other two immunization strategies, but is weaker than the new targeted immunization given in this paper.

Abstract： In order to solve the two problems of fixed threshold and offset constant, an improved non-parameter CUSUM(D-CUSUM) intrusion dentection algorithm for industry control systems which require high real-time and availability was proposed in this paper. Unlike to set a fixed value with previous experience, the algorithm redesigned the generation method of non parameter CUSUM offset constant β based on the theory of the famous Kolmogorov inequality. The CUSUM detection threshold τN was dynamically set using an alarm control parameter. Now, the β and τN have adaptive characteristics. In the simulation experiment of attacking temperature control system based on MATLAB Simulink, the results prove that the algorithm can improve the real-time and false alarm rate performance of detection and realize the real-time intrusion detection at a low false alarm rate.

Abstract： Mine uninterruptable power supply is important power supply equipment for the mine monitoring and communication system. A mine uninterruptable power supply with battery management system is introduced. Integrated switch regulator LM2576 is adopted to constitute DC/DC circuit and realize regulated output of the power supply. 20 Ni-MH batteries are used as backup batteries, low power 16-bit micro-controller MSP430F149 and special chip for battery group management LTC6803-3 are adopted to realize the battery status monitoring and management functions, which keeps voltage balance of the batteries. At the same time, the information collected such as the monomer voltage, current and temperature are displayed on the LCD12864. Test results show that the designed uninterruptable power supply for mine meets various parameters requirements of the system. The power supply is of simple structure, fully battery management function, safe and reliable operation. It has important practical significance and promotion value.

Abstract： A digital control system of phase-shifted full bridge DC/DC converter based on Infineon XMC4200, autonomous charging for AGV is studied in this thesis. The control system is designed according to the analysis of the main circuit topology of phase-shifted full bridge DC/DC converter and the characteristics of XMC4200. Small volume and high power efficiency are features of this system. Experimental results verify the validity of the theoretical analysis and feasibility of the digital control system.This converter could meet the requirements of the lithium-ion batteries in AGV applications which usually need charge with low voltage and high current frequently.

Abstract：The intelligent high power switching power source used in the telescope shaft drive power system is designed based on pulse width modulation(PWM) control chip SG2525 and IGBT module. The full-bridge inverter circuit and the full-wave rectifier circuit are used in the main circuit, in addition, the switching power is connected to the computer through an intelligent interface, at the same time, short circuit, over current, over temperature and lower voltage protection circuits greatly simplify the design of the system and improved the reliability of the system. The switching power′s control circuit is analysed and designed in detail, the output voltage can be adjusted continuously to satisfy some different specification motor. The validity of system is also verified by experiment .

Abstract： A kind of wireless charging system for portable electronic devices was designed on the basis of analyzing the theory of inductive wireless power transmission converter. It can automatically identify the target object and achieve the protection of output overvoltage.Integrated transmission of energy and signal was established between receiving and transmitting port in the system.Compared to the independent transmission mode,the power density of the system can be improved and hardware costs can be reduced.According to experiments, the influence on transmission efficiency caused by changeable load and distance between transmitter coil and the receiver coil was analyzed, which verifies the feasibility of the theoretical analysis.

Abstract：Collaborative filtering is the most widely used method in recommendation system. As the single computer is not suitable for the real-time and scalability of the recommendation system, a hierarchical collaborative filtering algorithm based on spark is proposed. First, according to user action sequences，a user interest model is created in the algorithm. Then, the parallelized Expectation-maximization(EM) clustering algorithm based on Resilient Distributed Datasets(RDD) is realized，and as a result, users is divided into different user cluster model. Last, the parallelized item-based collaborative filtering is achieved based on different user cluster. The experiments in Alibaba tianchi datasets show that the algorithm can significantly reduce the time, improve the accuracy of recommentation，and it has good scalability at the same time.

Abstract： For the intelligent maintenance of Auxiliary power unit is lower，auxiliary power unit fault intelligent diagnosis method based on BP neural network is put forward, and the mathematic principles and arithmetic are also expounded in this paper. Wavelet analysis and Butterworth digital filter are used in data processing, noise and interference are filter out by this digital filter. Furthermore, using the data of a certain auxiliary power unit as the sample data, a fault intelligent diagnosis model of the auxiliary power unit is built up based on BP neural network. Modeling and calculation are done with MATLAB，the results indicate that the fault intelligent diagnosis model not only has advantages of fast processing, noise suppression capability and so on, but also diagnoses accurately. This will improve the efficiency and the automation maintenance level of airlines.

Abstract：This paper presented a high speed data transmission system based on PCIE bus master DMA mode. An x1 channel PCIE system was set up using the FPGA chip from V5 series products produced by Xilinx company. Several transmission manners, such as single word reading and writing, DMA reading and writing were realized by the PCIE interface board, which was developed independently. The data can be displayed via GUI on host computers and chipscope, which can also be used to validate the correctness of the data. Experimental results show that the data transfer speed can be stable at around 400 MB/s.

Abstract： Aiming at abdomen CT images of different patients, the fast segmentation algorithm for liver lesions is presented. Firstly, by using contour map and region algorithm in combination, the initial contour of curve evolution is pre-extracted, which is suitable for the batch. In order to reduce the impact of spurious boundary and to converge to deeper boundary concaves, the improved G-S model based on adaptive curve can be finally presented to accurately fit the contour. Compared with other classical segmentation algorithms, not only is this method used to improve the precision of active contour model, it also proves to save considerable time. At the same time, noise jamming can be removed. Experiments show that the proposed methods could effectively extract the lesion region of image and meet the needs of reality.

Abstract： A design of spectrum analyser with adaptive time compensation is proposed by the study of time-interleaved ADC theory and digital down conversion FFT algorithm. Firstly, high-speed data are obtained by four-channel ADC sampling, frequency-domain cross-spectral is used to estimate time mismatch deviation and correcte delay deviation with Farrow structure′s fractional delay filter. Then sampling data is processed by digital down conversion, and FFT module is reused to obtain spectrum information of signal. Finaly, the results of FPGA experiment show that time-interleaving spectrum analysis can not only reflect the acquisition signal spectrum information accurately, but also decrease hardware resource expenses in relatively.

Abstract：A multi-criterion feature evaluation algorithm (MCFE-DSEC) is proposed, which takes the improved D-S theory by Murphy as the fusion rule. MCFE-DSEC aggregates a collection of different single criteria, making a comprehensive evaluation of features and removing redundant features to improve classification accuracy. The multi-criterion feature evaluation algorithm is applied to the high-speed train fault data. The obtained results show that compared with the Borda-Count method and single criteria, MCFE-DSEC method can evaluate the features at every speed more effectively and is more adaptable with higher accuracy.

Abstract：Aiming at the problems of occlusion, illumination and expression variations exist in the face recognition, a face recognition algorithm based on improved RPCA is proposed in this paper. It is based on the face of sparse error component to accurately determine the differences between face images. Firstly, the low-rank and sparse error component are obtained by low-rank recovery. Then the sparse peace sliding between two descriptors are defined to represent the characteristics of sparse error component. At last, the descriptions of the character of face images are combined to classify. The experimental results show that in illumination conditions and occluded areas random case, the proposed error image classify algorithm has superior recognition performance in dealing with face recognition problem such as occlusion, illumination, facial expression changes and so on.