摘要:An autonomous optical navigation scheme for probes landing on small body is presented.The image coordinates of three pre-selected fixation-points (FPs) are obtained by using an optical navigation camera.And the distance from probes to each FP is measured using a laser range finder.Relative position for probes is acquired from three image coordinates and the distance.The real-time orbit for probes is determined by using the extended Kalman filter.The effectiveness of the scheme is validated using digital simulations.And the influence of the navigation precision upon the small body model error for small body is analyzed by simulations,under the certain sensors precision circumstance.
摘要:Polycrystalline silicon (poly-Si) has been obtained using low-temperature (<500℃),nickel-based,metal-induced crystallization (MIC) of amorphous silicon.Because crystallization outside of the nickel-covered regions is not only possible but also commonly desired,the technique of metal-induced lateral crystallization (MILC) is often utilized.Based on studies on the crystallization kinetics and material characteristics,a unified mechanism is proposed for MIC both under and outside of the nickel coverage.While the material quality of MILC poly-Si is significantly better than that of solid-phase crystallized poly-Si,the performance of MILC poly-Si thin-film transistors (TFTs) is quite sensitive to and degraded by the inclusion of MIC/MILC and MILC/MILC interfaces in the device channel regions.When such interfaces are eliminated,excellent TFTs are obtained that can be used to realize high performance systems-on-panels,including sophisticated displays based on liquid crystals or organic light-emitting diodes.
YAN Li-ping, HUANG Ka-ma, LIU Chang-jun, CHEN Xing, ZHAO Xiang
Vol. 31, Issue 5, Pages: 667-670(2003)
摘要:A fluidized bed combined with microwave has been widely used in industry processing.Maxwell's equations of fluid combined with the heat transport equation have been solved by FDTD to simulate the interaction between microwave and flowing liquid and calculate the field and temperature distribution in the flowing liquid.Moreover,reflection and transmission in the fluidized bed based on waveguide are also investigated with respect to the velocity of liquid,the obliquity,and diameter of the pipe,through which the liquid flows.Numerical results are in good agreements with experimental ones.
摘要:An RF micromechanical CPW switch on the isolation polysilicon has been developed by using polymide sacrificial layer,and by dielectric isolating technology.The preliminary test results are as follows:the Coff and Con are 0.21pF,6.1pF respectively;the actuated voltage is 22V.The isolation is 35dB in the off state,and insertion loss is 3dB in the on state.The process is compatible with silicon IC,so is the basis of monolithically integrating RF micromechanical switch and IC for high reliability and low volume.
ZOU Bei-ji, ZHOU Hao-yu, LV Ge-feng, SUN Jia-guang
Vol. 31, Issue 5, Pages: 674-678(2003)
摘要:The image deformation often appears in the image processing system.So it is necessary to develop a highly efficient algorithm to get a good result of the deformation.But most of the existing deformation algorithms are based on Pixel-Fill-Method,and there are some obvious faults in these algorithms:(1) When the shape of the deformation region is complex,the efficiency of these algorithms is lower.(2) Some errors may occur when the pixels are mapped from the original image region to the destination image region since one target pixel may be relative to several source pixels.To solve these problems,a new deformation algorithm based on calculus is presented in this paper.In this algorithm a complex shape deformation region will be divided into several sub-regions firstly,and then each sub-region will be divided into many Small-Rectangles,at last a standard rectangle fill algorithm is applied to fill pixels in the corresponding rectangles.The experimental results from our algorithm show that there is a better effectof image deformation and a higher efficiency when using our algorithm.And our algorithm has been successfully applied to a human face beautifying system based on its real photo which was developed by the authors.
摘要:With the ever-increasing development of Internet,today's data-centric applications require storage platforms to possess high scalability.It causes a data-model-mismatch problem between applications and current storage systems.This paper designs a new network-attached object-based storage device (NAOSD),which provides storage interfaces for structured data to solve this problem.The usage of NAOSD in cluster storage environments is proposed,which introduces the uniform storage of data and meta-data as well as query-locating mechanism.The performance analysis shows that its access time increases with the system scale logarithmically,which is better than the conventional systems.The simulator of NAOSD have been implemented,and experiments show it improves the performance remarkably.
关键词:object based storage device;cluster storage;locating mechanism;Network-attached storage
摘要:Aiming at the devastating multiple-time-scale burstiness of VBR (Variable Bit Rate) video flows,an optimized algorithm characterizing renegotiation which is based on multilevel regulator and multiple-time-scale bandwidth allocation and flow admission is presented.Through the renegotiation between video source and edge node of the network,video flows are properly enveloped before entering the network.The proposed algorithm enhances the capability of control and prediction on video flows,which significantly attenuates the impact caused by burstiness on network.For small and moderate aggregated flows,it could attain a multiplexing gain approximating the gain of unrestricted VBR flows.
关键词:multiple time scale;flow control;renegotiation
摘要:The Artificial Immune System (AIS) inspired by the natural immune system is introduced.Based on exhaustive algorithm,a new algorithm is presented as detectors generating method in the negative selection model,including its design,performance analysis and experiment.Both mathematical analysis and experiment show that the algorithm runs in time linearly with the size of the detector set.The new algorithm can possibly generate a more powerful detector set than exhaustive algorithm.
摘要:The theory of artificial neural networks has been successfully applied to a wide variety of pattern recognition problems.Associative memories are focused and fuzzy morphological associative memories FMAM based on morphological and fuzzy operations are presented here.The properties of FMAM are greatly different from those of classical associative memories and fuzzy associative memories.FMAM's storage capacity and tolerance capability of erosive/dilative noise are analyzed.Autoassociative FMAM has the unlimited storage capacity,perfect recall guarantee,recall in one step and fuzzy explanation.The experimental results demonstrate the effective performance of autoassociative FMAM.
摘要:The definition of PMC equation model (that is equation description of PMC model)was first set up.PMC model is a kind of most common system-level fault model.Owing to the characteristics that traditional graph theory diagnosis algorithm couldn't be without "t-diagnosabe" and "believe the most" (that is supposing the number of fault processors less than half the number of all processors in the system),the concepts of "absolute fault base"etc.were defined,and the tool of "body grouping"was fully wielded,finally a concrete method——equation diagnosis algorithm was found,on seeking all consistent fault patterns for the class of models without supposing "t-diagnosable" or "believe the most".
摘要:Dynamic Bayesian networks are a representation for complex stochastic processes.How to learn structure of Dynamic Bayesian networks from data is a hot problem of research.An evolutionary algorithm is proposed.Fitness function based on expectation is presented to convert incomplete data to complete data utilizing current best dynamic Bayesian network of evolutionary process.Thus dynamic Bayesian networks can be learned by using two Bayesian networks,prior network and transition network,to reduce the computational complexity.Encoding is given,and genetic operators are designed which provides guarantee of convergence.Experimental results not only show this algorithm can be effectively used to learn Dynamic Bayesian networks structure from incomplete data sequences,but also illustrate the role of hidden variables and the influence of genetic control parameters on learned model.
摘要:Several neighboring pixels based prediction schemes were proposed by exploiting the statistical relationship between pixel distance and correlation.And an edge-adaptive predictive VQ scheme was proposed for images containing distinct edge characteristics.Simulations demonstrate the improved predicting accuracy and coding performance of the proposed predictive VQ algorithms for image coding applications constrained at a low computational complexity level.
关键词:predictive vector quantization;neighboring pixels based prediction;image coding
LIU Huan-yan, WANG Rong, FENG Jun, WANG Zhi-gong, XIONG Ming-zhen
Vol. 31, Issue 5, Pages: 711-713(2003)
摘要:A 10Gbit/s limiting amplifier is realized in a 2μm GaAs HBT technology.The Whole circuit consists of an input buffer,three similar amplifier cells,an output buffer for driving 50 Ohm transmission lines and a pair of feedback networks for offset cancellation.At a positive supply voltage of 3.3V and a negative supply voltage of -2V,the power dissipation is 500mW.The input dynamic range is about 38dB with a constant output voltage swing(300mVp-p).The chip area is 1.15 time 0.7mm2.
WANG You-lin, CHEN Zhen-yu, QIAN Chen, ZHU Ning-hua
Vol. 31, Issue 5, Pages: 714-716(2003)
摘要:The relation of calibration equations is investigated.It has been shown that for the reciprocal test fixtures,the four equations obtained from the measurement with a transmission standard can be used simultaneously in the calculation.Choices of the equations in the thru-short-match (TSM) and thru-reflect-line (TRL) methods are discussed.Experimental results show that the different choice of equations will lead to different solutions,and the calibration accuracy can be improved by utilizing the established relation of the calibration equations and properly choosing calibration equations.The proposed theory can be applied to the thru-open-match (TOM) and line-reflect-line (LRL) methods.
摘要:A dynamic routing and wavelength assignment algorithm in all-optical network with limited-range wavelength conversion was proposed.The correlation of different wavelengths on different links in all possible routes is described by wavelength weight.Attempts are made to assign wavelengths on each candidate route using shortest-path algorithm based on wavelength weight and choose the best as its route and wavelength assignment scheme.Numerical results obtained for NSFNET and 12-node ring show that our algorithm can greatly reduce the blocking probability and the number of converters needed under uniform traffic or traffic dominated by larger hop routes.
关键词:all-optical network;limited-range wavelength conversion;routing and wavelength assignment
摘要:Silica impurity in the high-purity water that is used for MOVPE(Metallorganic Vapor Phase Epitaxy) processes affects the quality of GaAs material dramatically.The principles and methods of EDI(Electrodeionizaion) to remove silica from high-purity water are introduced.Based on comparison of the efficiencies of silicon removal experiment by using various water fabrication equipments,it was determined that in order to obtain the best silicon removal efficiency the pH values should be controlled between 8.5 to 11.The silicon concentration in high-purity water could be decreased to below 0.5μg/L which satisfied the requirement needed for the fabrication of metallorganic sources and MOVPE process.
LI Xiao-ming, BAO Dong-xing, YU Ming-yan, YE Yi-zheng
Vol. 31, Issue 5, Pages: 724-727(2003)
摘要:A method to predict block conflicts in direct-mapped caches is proposed,which is based on the block replacement behavior.Accordingly,a cache scheme named Conflict Prediction(CP) Cache is presented.Its architecture contains a direct-mapped cache and a small fully-associative cache,and a conflict prediction table(CPT) which dynamically decides the allocation of fetched blocks from next memory hierarchy.SPEC95 Simulation results show that the performance of CP cache is always better than the traditional direct-mapped cache with twice the size of CP cache.With the same block size of 32 bytes,the average improvement of miss ratio of (8+1)kB CP cache with 8-entry CPT is about 12.2% over the traditional 16kB direct-mapped cache.By comparison with other similar architectures such as NTS cache and PCS cache,CP cache not only requires less hardware tradeoff and simpler control but also improves hit ratio and bus traffic.
摘要:According to the specific characteristics,digital DC-DC converters are easy to be integrated with other digital systems.There exist many digital control algorithms,and some complex algorithms need integrated ADCs,which enhance the difficulty of designing.Some simple schemes use single comparator as the feedback input,showing poor dynamic performance.A new digital control scheme for DC-DC converter with variable voltage-step feedback was presented and it is based on the Buck converter of no-ADC,single-comparator and constant voltage-step feedback.Because only single comparators or window comparators are used,this scheme is easy to be implemented.It can dynamically alter the duty-cycle increments according to the feedback input,which can effectively reduce the dynamic response time under sudden changes of the application environment without enhancing the complexity of the architecture notably.
DONG Tao, HOU Li-ya, ZHU Li, ZHANG Wei-yi, YANG Chao-chu
Vol. 31, Issue 5, Pages: 737-741(2003)
摘要:In the Study of performance of heat transfer in microchannel heat sinks used for cooling electronic chips, the available methods mostly assume microchannel sinks with uniform heat flux distribution.Finite element method, which can analyze microchannel heat sinks with non-uniform heat flux distribution,is presented.Compared with the results derived from existing methods for instance massive Calculated Fluid Dynamics calculations,in spite of arbitrary heat flux distribution occurred,heat transfer performance about microchannel heat sinks can be analyzed efficiently and more rapidly by the finite element method presented in this paper.A Galerkin finite element formulation has been used to predict the surface temperature,the fluid temperature and the total thermal resistance of the microchannel heat sinks with different heat flux distribution, finite element method is used to study the effect of geometrical parameters on the performance of the microchannel heat sinks at last.
关键词:micro system;microchannel;finite element method
JI Yang, LI Ying-yang, DENG Gang, HU Jian-dong, ZHANG Ping
Vol. 31, Issue 5, Pages: 742-746(2003)
摘要:Adaptive modulation has been widely adopted by some newly developed wireless packet networks such as WCDMA HSDPA (Wideband Code Division Multi-Access High Speed Downlink Packet Access) and HiperLAN/2,etc.With careful consideration on the attribute of link bandwidth uneven distribution with time slots in the adaptive modulation system,here a novel packet scheduling algorithm,Adaptive Differentiated Compensation Fair Queuing,is proposed.In this algorithm.An analytical scheme based on workload is used,and several sub-queues are involved.It can provide QoS guarantee to all backlogged flows,fair residual bandwidth sharing,and rational compensation mechanism to conquer the burst error in wireless environments.Besides the algorithm,in order to simulate the adaptive modulation system,a new channel model based on multi-state Markov chain is proposed in this paper.
摘要:In order to evaluate the running quality of network management system,a concept quality of connectivity (QoC) of network management system is given.Suggestions to establish the evaluation metrics for QoC were presented.A set of metrics as evaluation metrics for QoC was proposed.These metrics shows the running quality of network management system.A fuzzy evaluating method to evaluate the running quality of network management based on the QoC metrics is designed.An evaluation model based on different local telecom switch network management systems and relative evaluating result are described.
JING Ting, ZHAO Xiao-qun, LI Qi, JIA Zhi-cheng, HOU Lan-tian
Vol. 31, Issue 5, Pages: 751-755(2003)
摘要:Quasi-perfect binary array pairs,a new perfect signal was defined in this paper.It was laid on the basis of perfect binary array、quasi-perfect binary array and perfect binary array pairs.The size and the transform features of quasi-perfect binary array pairs were also discussed.And by analyzing its Fourier spectrum,some good results are got.Besides,the quasi-perfect binary array pairs of size 2 to 24 were searched out by computer.
关键词:information theory;perfect binary array pairs;quasi-perfect binary array;array pairs;signal theory
摘要:2-D Two-Scatterer Algorithm(TSA) is developed which is useful for further research about radar target identification based on the 2-D complex image analysis.The approach combines the 2-D phase unwrapping method with 1-D TSA.The simple simulation result with the influence of noise is presented for illustrating the application of the 2-D TSA.
摘要:A strategy to compensate PMD with linear chirped Fiber Bragg Gratings was proposed for the first time.The DGD induced by the diametric load on a linear fiber Bragg gratings had been calculated.And the feasibility of the proposed method is confirmed through the DGD measurement of a linear fiber Bragg gratings with diametric load on it.
关键词:polarization mode dispersion (PMD);differential group delay (DGD);poincare sphere
WANG Zu-yuan, ZHUANG Zhen-quan, HE Jin-song, WANG Xu-fa
Vol. 31, Issue 5, Pages: 765-768(2003)
摘要:An image retrieval method based on shape was proposed by using wavelet coefficients Gaussian density distribution.This approach forms the feature vector of the shape by taking advantage of the statistical attribution of wavelet transformation coefficients.Experimental results show that the extracted feature vector can depict the shape outline accurately,and remain invariant to the image's scale and rotation.It is an efficient image retrieval method based on the shape content.
关键词:wavelet transformation;generalized Gaussian density distribution;image retrieval
摘要:Based on the backward compatibility requisition,a new solution scheme of space-time trellis code for Wideband Code Division Multiple Access (WCDMA) system is proposed.Utilizing the space-time trellis code (STTC)proposed,WCDMA system has the same backward compatibility characteristics as space-time transmit diversity (STTD) does,and improved performance is also obtained compared with STTD.
摘要:The couplings of probes parallel to the major and minor axes in an elliptic waveguide are discussed by means of dyadic Green's function and reaction principle.The waveguide is semi-infinite.The reflection coefficient at terminal plane is Γ.The mutual impedance is given and plotted.
ZHANG Xiao, WU Min-qiang, YUN Shuang, LU Sang-lu, XIE Li
Vol. 31, Issue 5, Pages: 783-785(2003)
摘要:Continuous media caching design is a very critical problem.A caching strategy called EA caching algorithm was presented.This algorithm considers the requirements of serving existing and anticipated demands so as to improve the utilization of memory.The theoretical analysis and experimental simulation have proved that the performance of EA algorithm is far better than the traditional caching algorithm.
关键词:scontinuous media service;caching design;replacement algorithm
摘要:A method for airplane-like targets recognition is suggested,by using the position information of prominent point scattering,taken from the high resolution radar echoes by means of Relax algorithm,as the physic feature for recognition,which is based on the simplified model of points scattering under high resolution radar condition.Computer simulations with radar data show that for the trained data-segments that aircraft can be indicated at averaged rate of 96.77% to aspect-recognition and of 99.80% to class-recognition,respectively.For untrained data-segments the averaged class-recognition rate reaches 81.25% and 89.87%,respectively under two given conditions.
ZHANG Sheng-cai, ZHENG Yun-guang, GUO Hui, LI Shu-rong, ZHANG Shi-lin, GUO Wei-lian, HU Ze-jun, XIA Ke-jun
Vol. 31, Issue 5, Pages: 790-792(2003)
摘要:By adopting the approximate and simplified mathematical expressions and the relative equivalent circuit for the first time in consideration of the low injection effect,the high injection effect and the early effect etc.of the transistor,the mechanism for the origin of the negative resistance in photoelectric dual base transistor (PDUBAT) has been described.The theoretical calculation is well consistent with experimental results in the negative resistance region and the valley region.
关键词:photoelectric dual base transistor;feedback current;transient current
摘要:With the development of Internet and software reuse,it will be possible that end user who wants to customize some functional feature can get such program by finding some components from component libraries and compiling them according to the user's demand.A program mining methodology under networked environments is presented:with the aid of intelligent agents,analyzing and understanding user's requests for computing,identifying and searching component candidates from online component libraries,composing and reassembling them to form programs that perform the expected computing.The basic concepts,system framework,overall process and agent platform of program mining are investigated.The research issues are also discussed to present an open research area.
摘要:A new pseudo-random sequence-Phase Controlled(PC) sequence was introduced.The PC sequences have many good properties,i.e.very large linear span,good period cross/autocorrelation functions,and all sequences are balanced.Also we propose the method,by replacing the GMW sequence required in constructing PC sequence with the cascaded GMW sequences,thus can improve the properties of the PC sequence.The availability of this method is verified.The improved PC sequences have larger linear span than before,with the same properties of period cross/autocorrelation functions and balance as before.