In order to take advantage of the wireless sensor network to achieve more precise positioning of objects, the aims is to study and improve RSSI(Received Signal Strength Indicator) localization algorithm. According to the simulation and performance analysis on RSSI localization algorithm, the experiment determined the source of algorithm's errors. The influence of the number of neighbor bencon nodes on the positioning accuracy is discussed, and the appropriate number is selected according to the simulation figures and the data. Used the average value of RSSI to select the optimal value and eliminate the distortion data. The experi
mental results show that in the improved algorithm the localization error is reduced from 0~5 m to 0~1 m, and the average error is reduced from 1 meters down to
 0.1 m in the same environment, improving the localization accuracy.

To resolve the problem that the weak response of a deep target signal and the serious measuring environmental noise weaken the definition of the received signal greatly in the practical exploration of CSAMT(Controlled Source Audio-frequency Magnetotelluric), the cross-correlation algorithm based on the theory of the random signal statistics to deal with the noise of the CSAMT data is proposed. The method uses the fact that correlation between emission and received signals is strong, and the correlation between emission signals and random noise is weak, designing an emission signal current waveform recorder, dealing with emission signals and received signals by cross-correlation to filter out jamming signals, such as random noise etc. The experimental results show that the method can suppress noise and improve the measuring accuracy of CSAMT.

In order to solve the problem of locating underground personnel when the accident happens, a simple distributed vibration sensor system is proposed which employs an optical fiber ring interferometer having an additional sub-loop coupled with a main ring by a 3 dB coupler according to the advantages and disadvantages of communication system under mine. Comparing multiple responses of interfered beams by the lights travelling n times in the sub-loop, the proposed system can detect the position and the amplitude of acoustic vibration occurred in the main ring. By simplifying the calculation, the optical fiber vibration position obtained. The calculated values are compared with the actual value, it is showed that the system can detect the shock position very accurately with smaller error. When accident happened, miners strike the ground or wall to get vibration signal, it can detect the specific location of the miners providing a good basis for mine rescue.

As to the hidden terminal problem which means wrong result because of occlusion, we develop a cooperative spectrum detection system based on the proposed data screening technology. According to the relationship between the distance and the energy, we screen the detection information of effective cognitive users from all cognitive users to compose the best information collection. We judge the best information collection and get the results by combining the weight. The experimental results show that the detection performance of the proposed method is better than the ones of the traditional optimal weights merging algorithm and the perception algorithm based on signal to noise ratio comparison.

An electronic nose (enose) system has been designed in order to extend its application in the cosmetic field. The system is composed of four main parts including data acquisition, sensor signal conditioning, feature extraction and pattern recognition. The core of hardware used for data acquisition and
 transmission is dsPIC30F6014A. The sensor arrays are composed of three commercially available metal oxide semiconductor sensors by Figaro with an integrated temperature and humidity sensor. The PC control software of the e-nose system embedded with pattern recognition algorithm was developed by mixed programming between VC and Matlab based on COM component and made available as an executable file. Pattern recognition is based on SVM (Support Vector Machine), which enhances the generalization performance of small sample compared to other methods such as PCA(Principal Component Analysis), Neural Net and so on. The experimental results indicate that it is feasible to classify four kinds of perfume with up to 92% accuracy and identify the unknown sample using the e-nose system.

Considering the problem that the online solving speed of the advanced MPC(Model Predictive Control) algorithm is slow, we present a design method of MPC controller based on FPGA(Field Programmable Gate Array) to improve the computational speed of MPC controller. By analyzing the circuit, we design its fixed point model. The RTL(Register Transfer Level)level model of MPC controller is designed by using Verilog hardware description language and pipeline method, with designing Matrix addition, subtraction, multiplication and other computing modules in the bottom, MPC controller is ultimately implemented on FPGA. Finally, regarding the ETC(Electronic Throttle Control) as the controlled object, we verify the effectiveness of MPC controller based on FPGA through real-time simulation tests.

According to the control allocation problem of near space vehicles, a control allocation method based on HOFSA(Hybrid Optimization Fish Swarm Algorithm) is studied. HOFSA fused by differential evolution and genetic evolution is applied to dynamic constrained control allocation problem by fully considering the dynamic constraints. The initial fishes are generated randomly according to the constraints of the control surfaces, and the fish swarm algorithm is used to achieve a global search. The fish swarm algorithm can converge rapidly in the initial stage, but the convergence speed will be slow down or keep stagnant when search local region. Due to its advantage in local optimization, the differential evolution algorithm is further employed to optimize for improving convergence speed and accuracy. And under the circumstance that differential evolution has not been able to get more optimal solution, the genetic algorithm is used to avoid the local optimal solution and to improve the efficiency of the whole allocation algorithm. The developed control allocation method based on HOFSA is applied to a near space vehicle and the simulation results show that it can effectively assign control instructions to each control surface and the tacking control performance is good.

Focusing on improving the objectivity and accuracy of AHP(Analytic Hierarchy Process) in the data quality evaluation, improving to each level indexes assignment method of AHP, we use weighted quadrature relationship instead of weighted sum relationship between the higher index and the lower index, introduce an objective assignment method of expert scoring, assign the weights on the indexes matrix in the quality evaluation of 4G geological spatial integrated data. The comprehensive weights results of indexes matrix and the evaluation results are more objective reflect the quality of the integrated data compared to the results by typical AHP, and can provide the basis and quantitative reference for data quality control of G⁴I system.

Aiming at the trajectory tracking problem of rotating wing micro aerial vehicle, an adaptive sliding mode controller based on adjusting the PID（Proportion Integration Differentiation）gain is proposed. To achieve the trajectory tracking error robustness of the model parameter variations and external disturbances, basing on Lyapunov stability theory, an adaptive law is adopted to adjust PID control gain parameters online. Saturation function is designed to inhibit sliding mode controller high-frequency chattering. Simulation results illustrate that the designed controller can drive the micro aerial vehicle to track the desired trajectory accurately, obtaining high precision, and can meet the requirements of practical engineering applications.

In order to resolve the problem that the cable USB(Universal Serial Bus) devices are placed close and chaotic, we design a wireless usb device. The device can be used as a pointtopoint device or a point-to-multipoint device, which can achieve the function of a wireless printer. The system is based on S3C2440 ARM9 processor and Linux operaion system. The wireless communication is based on the WiFi(Wireless Fidelity) technology. The experiment results showed that the device could achieve wireless-printing function well.

In laser beam quality measurement, in order to solve the problem of highprecision control of laser intensity, we proposed an adaptive attenuation method for laser light intensity adjustment. The method is based on microcontroller adaptive control algorithm. The laser intensity precision attenuation ratio is calculated using the spot image gray level PC returned. A motor controls two attenuators to adaptively change laser light intensity, achieving a closed-loop system quickly and accurately which obtains the specific light intensity laser spot image. Compared with previous attenuation control methods, the experiment shows that this method can accurately obtain high-precision gray value laser spot image, and the single measurement of the laser beam quality spend only 5 seconds, and system measurement only 3 minutes.

Traditional artificial security patrol system is restricted by human factors, the patrol route, time, frequency, and the density is not reasonable. We design a car which has multisensor such as flame sensor, ultrasonic sensor, light sensor and so on. It uses freescale main controller and auxiliary controller which coordinate with each other to complete tracking and obstacle avoidance, low battery detection, temperature acquisition, smoke alarm, video transmission, intelligent control of motor speed and accurate positioning and so on. Patrol car uses PC for remote control and uses the real-time transmission image to test the security target. From the experiment test we know that the cars maximum error is 3%, so its intelligent patrol effect can meet the development requirements of the current requirement, and the small form, full functions and strong practicability all shows that it has a broad market application prospect.

To improve the performance of DV-Hop(Distance Vector-Hop) algorithm in random WSN(Wireless Sensor Networks), a modified location algorithm based on hop thresh and node classification is proposed. The algorithm chooses farther beacons to estimate the average distance of each hop, then the unknown nodes classify the beacons according to the hop and compute weighted and modified distance to beacons. The simulation researches the impact of node density, beacon ratio and communication radius to algorithm performance, the results show that the algorithm can get better estimates of the hop distance, which can improve location precision and reduce the maximum location error in random network.

The security mechanism of DaaS (Database as a service) is mainly addressed. DaaS can solve the data security problem using encryption method. A new mo
del based on NetDB2-MAMS (NetDB2-Minimum Attribute Multi-Secret Sharing) is presented, user data can be protected by getting minimum shared attribute set and using secret sharing technology. Experiments show that the model can protect the privacy of DaaS and avoid the cost of encryption and decryption.

To solve the problem that the environment changing imparts the system model, a good way to improve the performance of the controller is to use multiple models control if models are approximately available for different environments. We design a PID(Proportion\|Integral\|Differential) performance index with weighting factors and forgetting factors, which make the performance index more in real-time. This backstepping controller applied to Robot Manipulator, differs from standard backstepping in that the virtual control inputs are designed based on estimates of the previous virtual control inputs. This eliminates the need for taking derivatives of the system dynamics, which simplifies the control law. The design methodology is based on the Lyapunov stability. The simulation results demonstrate the effectiveness and fasibility of the proposed control strategy. PID-type switching index function, which increased forgetting factor making the witching more accurate and reasonable. The simulation results demonstrate the effectiveness and fasibility of the proposed control strategy.

The finitetime synchronization control problem is addressed for complex dynamical networks with linear couplings. In the case that nonlinear dynamics satisfies certain conditions, based on the synchronization error, two different continuous nonsmooth control protocols are proposed, combined with linear feedback principle and finite-time control. The sufficient conditions are given for finite-time synchronization of complex networks by applying matrix theory and finite-time
stability lemma. And finite time synchronization can be achieved with suitable control gain for complex dynamic network with nonlinear couplings. Simulation res
ults verified the correctness and validity of the proposed scheme.

In order to improve the efficiency of communication between robots, Zigbee wireless sensor network communication interface technology and robotics technology are combined together to solve the problem of multi-robot cooperative communication. A robot based on Zigbee wireless communication system is described, and this system can communicate with each other in addition to robots, enables real-time wireless communication system to send and receive information to build robot wireless communication platform. Experiment shows that normal communication between the wireless communication system improves the robots signal transmission problems and improves the transmission efficiency and practicality.

To overcome corner clustering phenomenon of Harris operator feature point matching, we proposed a probability density\|based corner matching algorithm. This method sets the image distance between the corner as a basic reference coefficient of the main regional division, using the corner probability density of regional division to reduce the matching area, and judging the feature points outside the region as false corners and removing them. Experiments show that the matching results of improved algorithm effectively reduce interference points, improving the timeliness and accuracy of the algorithm.

SPL (SoftwareProduct Line) is the hotspot of the research of software reuse, this paper in view of the defects of present SPL, analyzes these defects and gives the countermeasures. A process model and a seamless migration model of SPL based on software library are put forward, and a car\|embedded multi\|media system(simulation) are demonstrated based on the  process model. Relying on the  functions of  storage and retrieval of relational database. Application of the two models is conducive to reduce initial investment and contribute to the migration of SPL, and has a certain practical value and popularization value.

For relieving the pressure of mobile communication pivotal power resource, resolving the obsolescence of the hardware equipment, and satisfying the abstraction, transmission and loading of the quantities of business and data, the paper reforms the cloud computing architecture of data marts by the means of distributed computing and virtual technology and ETL（ExtractionTransformationLoading）tools of cloud computing. Through test and application, it showed that the reformed data marts in BASS（Business Analysis Support System）system become finer in data processing and become more balanced,stable,and high eff
icient in data exchange. And the data marts reduce the cost and increase the efficiency of the basic sources.