摘要:One of the key issues in the design of a QoS-oriented service is how to identify a feasible route which satisfies multiple constraints while simultaneously achieving efficient utilization of network resource.The underlying problem can be stated as a Multiple Constraints Optimal Path (MCOP) problem.Recent years,some algorithms have been proposed in literature,but suffered from various disadvantages.They can not meet the practical requirements.In this paper,a formal definition of MCOP problem is stated.A novel enhanced heuristic for solving MCOP problem is proposed,in which elegant path searching techniques are used to achieve higher performance in terms of path selection success ratio and average path cost.Using extensive simulations on random graphs and random assigned link weights,the improvement of the new algorithm proposed is verified,and the impact of some key design parameters on the performance is discussed.
摘要:In the actual OFDM system,the intersymbol interference and interchannel interference can be a dominant factor which degrades the system performance.This paper analyzes how ISI and ICI are produced without inserting CP.Then a simple method of cancelling ICI which can be done in frequency domain based on this theory is proposed.As to the equalization methods discussed in the paper,simulation results are provided which show that in Broadband Wireless LAN bit-error rate (BER) is distinctly reduced and it is applied easily.
关键词:Orthogonal Frequency Division Multiplexing(OFDM);frequency-domain equalization;intersymbol interference(ISI);interchannel interference(ICI);channel impulse response(CIR)
摘要:In this paper,we proposed a novel wireless communications architecture:distributed virtual cellular network(DVCN).In DVCN,virtual cell (VC) can remove and change area freely,which brings some benefits of capacity,wireless resource control and so on.Then without considering the shadowing and multipath fading,we evaluated the reverse link capacity of VC and compared it with distributed antenna (DA) system.The results show DVCN can reduce users'transmit power,decrease multi-user interference and improve the reverse link capacity.
摘要:In this paper,we present an amicable-orthogonal-design-based differential space-time block code (ADSTBC) for multiple antennas.Compared with the existing differential space-time modulation designs,our scheme imposes no restrictions on underlying signal constellation,and therefore can improve the spectral efficiency by exploiting efficient modulation techniques such as QAM,APSK etc.We then derive a maximum-likelihood differential decoder (MLDD) with linear complexity for flat Rayleigh fading channels.Particularly,when the QAM constellation is used in ADSTBC,MLDD can be further simplified to independently detect the real and imaginary parts of each data symbol,and accordingly the implementation cost can be greatly reduced;Moreover,the computational amount for detecting single data symbol is kept constant in MLDD while the order of the QAM constellation increases.
LIN Yu, WU Hai-tao, CHENG Shi-duan, WANG Chong-gang, JIN Yue-hui, WANG Wen-dong
Vol. 32, Issue 4, Pages: 552-556(2004)
摘要:The packet pair/train technique estimates the capacity bottleneck of a path from the dispersion experienced by two/multiple back-to-back packets.This paper analyzes the measured noise and explains the distributions of measured samples under different probing packet sizes and different packet train lengths.Based on the noise analysis and a 'signal model’,a new algorithm called MBFA (Model-Based Filtering Algorithm) is presented,which uses 'signal model’ to enlarge the signal and to eliminate the noise.
摘要:Space-time codes exploit the presence of multiple transmit and receive antennas to improve communication performance on multipath radio channels.Increasing the number of antennas increases the required training interval for channel estimation and reduces the available time in which data may be transmitted.In this paper,by combining the subspace method with the special structure of space-time block code,we propose a blind channel estimation algorithm,which does not require any training symbol.With the special structure of space-time block code,in the received signal,the vectors of channel fading matrix have some relation with each other.Combining this relation with subspace method,we can accomplish the blind channel estimation.Monte Carlo simulations show that when the SNR is low,the loss in decoding performance incurred by the error of channel estimate is relatively small.
关键词:space-time block codes;blind channel estimation;subspace method;array signal processing
摘要:This paper puts forward an adaptive localization method that can resolve switch point localization problem in the partial backward decodable bit stream (PBDBS),e.g.bit position method.And we compare this method with the H.263 codec and the original PBDBS method.It has been found that this method can obtain better error resilient performance and can be applied in video transmission over Public Switch Telephone Network (PSTN) and wireless channel.
关键词:PBDBS;error resilient coding;error concealment;robust video transmission;H.263
摘要:A weighted algorithm of sparse decomposition is developed for recovery of signal in strong background noise.To find the real components in a complete dictionary,the cost function can be constructed by a weighted sum of the l-2 norm of residual errors and l-1 norm of sparse components.Taking complete dictionary as the multiresolution wavelets,a feasible penalty formula is deduced according to two-scale relation of additive noise in wavelets dictionary.Analyzing the resolving process of minimum l-1 problem,proposed is the difference of l-1 norm of signal components as converge condition,where the difference is derived from the results of successive two iterative steps.The method is confirmed by both simulated and real data.
摘要:During the course of monitoring and controlling on industrial process,because physical process changes instantaneously,the optical CT can not take "static" imaging like the medical CT while it must carry out "dynamic" imaging (that is namely flow imaging),which would lead to not enough projection angles and the number of projections for light beam scanning.According to this,the reconstruction algorithms such as Linear Back Projection (LPB),Filter Back Projection (FBP) and projection rearrangement method cannot be applied directly and must be improved.So,in this paper,the application feasibility of fan-beam projection rearrangement method in optical flow imaging is taken into account,the calculation approach is analyzed,and the restriction relations among the number of light sources,number of detectors and fan beam-divergence angle,what interpolation precision affects on image quality is analyzed too.In simulation experiment,by comparing the reconstruction images by LBP and projection rearrangement method,the image quality and real-time performance of the method in this paper are estimated.
YU Heng-yong, MOU Xuan-qin, TANG Shao-jie, CAI Yuan-long
Vol. 32, Issue 4, Pages: 575-579(2004)
摘要:Three-dimensional (3D) reconstruction of blood vessel based on rotational digital subtraction angiography (DSA) is a new hot issue in the field of medical image processing.And it has a prospect of being widely used.On the basis of simultaneous algebraic reconstruction technique (SART),the state transition probability is constructed depending on a weighted average of all rays that traverse the voxel,thus an iterative reconstruction algorithm for binary object is developed.According to the character of 3D binary blood vessel,the maximum uniform criterion is adopted to constrain the iterative procedure.It has a function of volumetric cluster and it can greatly improve the reconstruction quality.The simulation experiments are executed on Defrise model and coronary artery model.The experimental results show that the proposed algorithm in this paper is better than Feldkamp algorithm especially in the aspect of suppressing noise and preserving structure information of reconstructed object.
摘要:In this paper,an algorithm which automatically transforms number to gray-image using the number as key is proposed,distinct number leading to distinct gray-image.Its implementation is that random sequence is made by chaotic iterations using number as seed,then the sequence is transformed to two-value image.By cellular automata processing and after smoothing operation the gray-image is generated.The gray-image is regarded as a digital watermarking pattern and the pattern has most of its power concentrated in low frequencies.The watermarking is embbed into original image in spatial apace,and it is detected by using DCT.The experiment results show that the pattern has good property.The watermarking algorithm based on the pattern can resist some usual attacks.
摘要:Support vector machine (SVM) is a new statistical learning method.Compared with the classical machine learning methods,the learning discipline of SVM is to minimize the structural risk instead of empirical risk used in the learning discipline of classical methods,and SVM gives better generative performance.Because SVM algorithm is a convex quadratic optimization problem,the local optimal solution is certainly the global optimal one.In this paper,SVM algorithm is applied to detect the micro-calcifications in mammogram for the first time.The algorithm is tested with mammograms of clinical patients and results show that SVM method achieves a higher true positive in comparison with artificial neural network (ANN) based on the empirical risk minimization,and is valuable for application in clinical engineering.
摘要:Not using phase-difference methods or phase correlation methods which are popular to gain the disparity map in phase-based stereo,the approach presented in this paper directly uses the fact that the phase structures are equal at the correspondence locations.A cost function is defined to measure the matching error between point x in the left view and the one shifted by the assumed disparity value in the right view as the aggregation of the phase differences between their local windows.To complete a globally optimized search process,dynamic programming technique is employed to find the disparity results which minimize the differences between the local structures in two views.In addition,this approach automatically tackles the singularity problem and the occlusion detection problem in the optimal disparity search process,instead of treating them as secondary processes that are postponed until matching is completed.The experiment results demonstrate that this approach has good applicability.
摘要:The paper extends the original Alternating Projection Neural Network(APNN) and proposes an Extended Alternating Projection Neural Network (EAPNN) which functions in the field of complex numbers.An improved weight-learning approach,which permits linear dependence of complex patterns,has been presented.A general mathematical expression of the EAPNN steady-state solution has been obtained.From the general mathematical expression we have derived the sufficient and necessary condition of the EAPNN configured as a content addressable memory.In addition,simulation experiments have been designed to verify the theoretical analyses in the paper.Finally it is pointed out that the EAPNN has been applied to signal processing such as band-limited signal extrapolation,notch filter and weak-signal separation.
关键词:alternating projection;neural networks;content addressable memory;condition of convergence
摘要:Resources are the bases of human life.Energy is the characteristic resource in Industrial Age and therefore,the Law of Energy Transform and Conservation became the solid foundation of science and technology in modern time.Information is the characteristic resource in Information Age and thus the Transformational and Unified Theory of Information,Knowledge,Strategy and its Execution will be the soul of science and technology development in Information Age.The concepts and the mechanism of information,knowledge,strategy and its execution will be explained and the unified theory of information,knowledge,strategy and its execution can thus be established within which the information theory has yet greatly gone beyond Shannon Theory,the knowledge theory has also dramatically extended from knowledge engineering,and the three existing approaches of intelligence theory,the neural networks approach,the expert systems approach,and the sensor-motor approach,are integrated into a uniform system.The unified theory will provide new vision and new vigor for the further development of intelligence science.
ZHAO Li, JIANG Chun-hui, ZOU Cai-rong, WU Zhen-yang
Vol. 32, Issue 4, Pages: 606-609(2004)
摘要:This paper presents a new emotion recognition method based on the hybrid feature and MMD.The hybrid feature was composed of global and time sequence.Experiments were conducted on a task of 10 speakers' 1000 sentences including happy,angry,surprising and sorrowful emotions to demonstrate has effectiveness of the new method.The average emotion recognition rate was reached as higher as 94%.
摘要:The modeling of topological relations between spatial regions is an important research topic in spatial reasoning,Geographic Information Systems (GIS) and computer vision.The modeling of topological relations between indeterminate regions has attracted plenty of attentions from researchers during the recent years.An extended egg-yolk model between indeterminate regions is proposed based on three predications.The model is cognitively valid,and can uniformly deal with crisp regions as a specific case.It extends the original egg-yolk model based on RCC5 and RCC8 relations and can analyze topological relations on multiple levels.
摘要:We have discussed about how to extract signal in order according to signals'character in this paper,which is an important problem in the subject of blind separation.A sequential blind signal extraction algorithm in order based on genetic algorithm is proposed,which can ensure extraction of source signals accroding to the order of absolute kurtosis of signals,it avoid the problem that many algorithms based on gradient descent approach get into part extremum easily.Moreover,a deflation algorithm based on Schmidt orthogonal is proposed,which not only simplify the computation of the deflation algorithm in Cichocki- Thawonmas-Amari(1997),but also whiten the mixed signals in the same time.Simulation results show that the algorithm can ensure the extraction of signals in order.
关键词:genetic algorithm;sequential blind extraction in order;kurtosis;schmidt orthongnal
摘要:The accuracy of the frequency estimation technique of sinusoid based on the interpolated FFT is investigated.Formulas for evaluating the root-mean-square (RMS) error of frequency estimation for different signal-to-noise ratio and different FFT length without weighting window and with Hanning window are derived.Study results show that when the frequency of the signal is close to the center of the main lobe of the FFT,due to the inversion of the interpolation direction,the interpolated FFT without weighting window results in large errors in frequency estimation.Therefore in noise background,the frequency estimation error of the interpolated FFT method is much larger than that achieved by Jain in [2] using a given pure test sinusoid signal.The effect of the weighting window on frequency estimation accuracy is also discussed.Computer Monte Carlo simulation results fit well with the formulas given in this paper.
摘要:The paper surveys image mosaics technology.Image mosaics are the technologies that a large,seamless and high-resolution image is made from a set of images.The paper calls the 2-D surface formed by the image mosaics as the manifold.According to the mode how the manifold is formed,it divides the image mosaics into two kinds:adaptive manifold-based image mosaics and man-selected manifold based one.Adaptive manifold-based image mosaics are produced by the strips from video.Man-indicated manifold-based image mosaics are made from image alignment.It includes local alignment,global alignments and image blend.The paper classifies and discusses the above technologies.At last,it gives an evaluation and discusses the future development of image mosaics.
摘要:The paper considers the static shared-path protection problem under SRLG (Shared Risk Link Groups) constraint in meshed WDM networks-determining the backup route for each wavelength demand to minimize spare capacity required,given the network topology,traffic matrix and primary routes of all demand.This problem can be formulated as Integer Linear Programming,a NP-C problem,which is computationally intractable for large size of networks.So,we present a heuristic algorithm to address the problem.The key idea of the algorithm is to maximally share the capacity.Simulation results show that the developed scheme effectively reduces the required spare capacity and enhances the network resources utilization.
摘要:It's proved that the nonzero linear combinations of the coordinates (NLCC for short) of a bijective monomial in a finite field of characteristic two are linearly equivalent,and the number of equivalent transformations is equal to the number of nonzero elements in the finite field.It's prove that the NLCCs of S-boxes of AES are linear equivalent,and the group formed by the zero transformation and all transformations constructed in this paper for equivalence of a given NLCC to NLCCs under the pointwise addition of transformations is isomorphic to additive group of the finite field.The equivalent transformations of the least significant coordinate to 8 coordinates are given,which is a base of this group.It's proved also that the sum of equivalent transformations of coordinates constructed by Fuller J and Millan W is not an equivalent transformation of two NLCCs of S-boxes of AES.
摘要:When a group of people want to communicate securely over an open network,they need to run a conference-key establishment protocol to establish a common conference key K such that all their communication thereafter are encrypted with the key K.By using the secret sharing scheme based on the MDS code as the basic component in this paper,we propose a new provably secure conference-key agreement protocol under the broadcast channel.In the protocol,a passive adversary gets no information about the conference-key established by the honest participants under the assumption of general Diffie-Hellman problem;and the honest participants can agree on a common conference-key no matter how many participants are malicious.
FENG Yan-qiu, CHEN Wu-fan, LIANG Bin, LIN Ya-zhong
Vol. 32, Issue 4, Pages: 645-647(2004)
摘要:Fuzzy c-means(FCM) clustering is one of well-known unsupervised clustering techniques,which has been widely used in automated image segmentation.However,when the classical FCM algorithm is used for image segmentation,no spatial information is taken into account.This causes the FCM algorithm to work only on well-defined images with low level of noise;unfortunately,this is not often the case in reality.In order to overcome this limitation of FCM,the prior spatial constraint is incorporated based on Gibbs random field theory.The definition of refusable level is presented and then new clustering object function is presented.This new algorithm connects Gibbs random field with FCM algorithm and is shown to be most effective in our experiments.
关键词:image segmentation;FCM;Gibbs random field(GRF);multilevel logistic model (MLL)
摘要:The binary sequence generator with memory is vulnerable to "linear sequential circuit approximation" attack though it can resist traditional correlation attack.With the idea of table-shuffling,a kind of binary sequence generators based on time-varying binary combiner are given.Some cryptographic properties like correlation between input and output are discussed.Several known sequence generators are of this model.The model can be useful in designing sequence generators.
摘要:This paper gives a proof that all Bent functions of degree 2 are linearly equivalent each other and presents a method for constructing multi—dimension Bent functions when one output variable is of degree 2.For a given reversible symmetrical matrix with the elements of its diagonal is 0,it provides a method for constructing k-1's reversible symmetrical matrixes,such that every non-zero linear combination of these k's matrixes is also a reversible symmetrical matrix.
关键词:Bent function;multi-dimension Bent function;The families of Bent complementary functions;linear equivalent;reversible symmetrical matrix
摘要:The research on digital watermark is a focus of multimedia signal's processing.So far,there have been many methods to embed and extract watermark.What is important to digital watermark is how to improve its robustness.In this paper,the independent component analysis (ICA) is used in our watermark scheme,that is,we mix the host image and the watermark directly in spatial domain and then use ICA to extract the watermark.We get excellent simulated results.The simulated results show that it is a new method with very outstanding robustness.
摘要:Image segmentation is an important research topic in the field of computer vision.A histogram-based multi-threshold automatic segmentation method is presented for gray scale images.On the one hand,it uses the weighting fuzzy c-mean clustering algorithm to fast implement the segmentation procedure;on the other hand,it automatically determines the proper number of threshold via unimode statistical test.The experimental result illustrates the effectiveness of the proposed method.
关键词:image segmentation;cluster analysis;weighting fuzzy c-means algorithm;statistical test
摘要:Performance degradation due to Inter-Carrier Interference (ICI) resulted from channel Doppler effects is significant for Orthogonal Frequency Division Multiplexing (OFDM) when used in mobile communication systems.In this paper,the mechanism of OFDM ICI caused by Doppler effects is discussed.Based on the I/O relationship of channel output Doppler-spread function,a 3-dimensional graphic form for visualizing the received OFDM signal structure and a simple method for calculating ICI levels of arbitrary subcarrier in channels,esp.with a continuum of multipath components,are proposed.
摘要:Based on ICA (Independent Component Analylsis) principle,a fast leaning algorithm for blind signal separation is presented.By seeking the local extrema of the fourth-order cumulants (i.e.kurtosis coefficients) of a linear combination of the observed variables,the model and the process of this algorithm are obtained and then used for the experiment of blind signal separation.The results of the experiment show that this algorithm has a great deal advantages in blind signal separation and features extraction such as fast convergence and needless in any dynamic parameter and the like.The algorithm can separates all independent components from blind source signals,which is non-Gaussian distribution.The algorithm is a new,highly efficient and reliable method in signal processing.
关键词:blind signal separation;independent component analysis;kurtosis;neural algorithm
摘要:Based on the analysis of human visual system characteristics to brightness and color,we present a novel algorithm with considering about wavelet decomposition and color space for color image enhancement.The uniform color space CIE La b is used and a hybrid contrast enhancement algorithm for color image is proposed.Color image enhancement exists two goals:image looking naturally and looking vividly.Experimental results confirm that improved color image enhancement is achieved.
关键词:color image;wavelet transform;human visual properties;color space
ZHAO Shu-guang, LIU Gui-xi, WANG Jun-ning, YANG Wan-hai
Vol. 32, Issue 4, Pages: 680-683(2004)
摘要:To implement evolutionary design of analog circuits with larger scale and higher speed,we proposed a multi-stage adaptive Genetic Algorithm.It features an encoding-decoding scheme supporting creation of circuit structures and standard values (i.e.discrete values in common use) of components,a multi-objective fitness evaluation method based on PSpice simulation and penalty for the circuit size,and a genetic parameter adaptation technique considering the different effect of each locus on fitness and tracking the course of evolution.It is suggested by the experiment results of a series of active filter that the approach discussed can firstly search feasible circuit structures,then find an optimal combination of component values,and finally provide a simplified circuit with a response very close to the design target,whilst only a modest computational resource that a PC can provides is required.
关键词:evolutionary design of circuits;genetic algorithm;evolvable hardware
摘要:This paper presents a new method based on the distance transform and non-linear programming in 3D medical image registration,especially in the study of the object skin matching.In comparison with known methods,a simple and realizable 3D-distance transform and optimized algorithm are used.In the realization of actual matching experiments,we use variable ending conditions.When the searching points congregate on one point,we add the spans in all directions respectively in the matching search until a more optimized result is archived.Then the new search continues.The results show that it has important value in noise interference,local registration and so on in 3D-medical image registration.
KONG Xiang-yu, HAN Chong-zhao, WEI Rui-xuan, MA Hong-guang
Vol. 32, Issue 4, Pages: 687-689(2004)
摘要:In this paper,the decoupling problem for Volterra adaptive filters is researched,and fully decoupled RLS adaptive Volterra filters is presented.According to the pseudo-linear combination structure of Volterra filters,by applying the principle of RLS filter and constrained optimization theory,a fully decoupled Volterra normal equation with block diagonal input correlation matrix satisfying the minimum cumulative time series error criterion is educed.A fully decoupled RLS adaptive Volterra filter is then designed and a suit of adaptive regulating formulations for the weight vectors of the filter is given out.Simulation results indicate that the proposed method in the paper is efficient.
摘要:This paper presents a new differential state estimator,which does not rely on the model of the estimated system and has higher accuracy with a few parameters.Its parameters are distributed through root-locus and pole requirements.Stability and convergence of the estimator are proved.Frequency characteristic analysis shows its high quality filtering property. Further more,based on the estimator a high order differential feedback controller is designed,which does not rely on the model of the nonlinear system,and has well stability and robustness.Lastly,to illustrate,two simulation examples are given.
关键词:stability;convergence;differential state estimator;high order differential feedback controller;robustness
摘要:In this paper,the properties of rough approximate operator for bisection method of pattern are discussed.The possibility and necessity of pattern classification by dividing pattern space into two decision areas using the given patterns are described.The rough neural networks algorithm of the possibility and necessity of the classification problem with boundary areas is designed.Finally,numerical examples are given to illustrate the efficiency of algorithm.
关键词:rough approximate operator;pattern classification;possibility and necessity