Email alert
2013 Vol. 35, No. 4
Display Method:
2013, 35(4): 763-769.
doi: 10.3724/SP.J.1146.2012.01225
Abstract:
The particle filter is a nonlinear filtering algorithm and it is widely used in target tracking. In the basic particle filter, the particles are scattered by Gaussian distributions, and the target for low-speed movement and gentle changes can be better captured. However, when the motion status of target suddenly changed, such as sudden acceleration, deceleration, or turning, etc., it often leads to the target loss. The reason is that the tailing of the Gaussian distribution curve is relatively lighter than that of the Rayleigh distribution, and then it causes the distributing range of particles is too small to capture the target. Considering this issue, the Rayleigh distribution is proposed to spread the particles, and then achieve the particles reproduction and dissemination, the Rayleigh distribution parameters is adaptively adjusted according to the rate of movement of the target, and through the secondary particle filter method to improve tracking accuracy. The experiments demonstrate that the particle propagation mechanism based on Rayleigh distribution can effectively enhance the ability of capturing target.
The particle filter is a nonlinear filtering algorithm and it is widely used in target tracking. In the basic particle filter, the particles are scattered by Gaussian distributions, and the target for low-speed movement and gentle changes can be better captured. However, when the motion status of target suddenly changed, such as sudden acceleration, deceleration, or turning, etc., it often leads to the target loss. The reason is that the tailing of the Gaussian distribution curve is relatively lighter than that of the Rayleigh distribution, and then it causes the distributing range of particles is too small to capture the target. Considering this issue, the Rayleigh distribution is proposed to spread the particles, and then achieve the particles reproduction and dissemination, the Rayleigh distribution parameters is adaptively adjusted according to the rate of movement of the target, and through the secondary particle filter method to improve tracking accuracy. The experiments demonstrate that the particle propagation mechanism based on Rayleigh distribution can effectively enhance the ability of capturing target.
2013, 35(4): 770-776.
doi: 10.3724/SP.J.1146.2012.01095
Abstract:
In order to solve the problems of color similar target, scale change and occlusion in target tracking, this paper proposes an adaptive fragments-based target tracking method fusing color histogram and SIFT features. Taking color projection and imaging angle as the standard, the adaptive fragment method can ensure not only the distinctive features of fragments but also the number. Each fragment is represented by color histogram and SIFT features. Adaptive scale selection can be obtained by computing the scale change of SIFT features. The weighting factor and corresponding template of each fragment are updated during tracking and the whole target template will be divided again if the target appearance changes significantly. Experimental results demonstrate that the method can track the target accurately and effectively and has advantage in color similar target tracking, adaptive scale selection and occlusion handling.
In order to solve the problems of color similar target, scale change and occlusion in target tracking, this paper proposes an adaptive fragments-based target tracking method fusing color histogram and SIFT features. Taking color projection and imaging angle as the standard, the adaptive fragment method can ensure not only the distinctive features of fragments but also the number. Each fragment is represented by color histogram and SIFT features. Adaptive scale selection can be obtained by computing the scale change of SIFT features. The weighting factor and corresponding template of each fragment are updated during tracking and the whole target template will be divided again if the target appearance changes significantly. Experimental results demonstrate that the method can track the target accurately and effectively and has advantage in color similar target tracking, adaptive scale selection and occlusion handling.
2013, 35(4): 777-783.
doi: 10.3724/SP.J.1146.2012.01023
Abstract:
This paper presents a master-salve tracking framework for wide area video surveillance with two Pan-Tilt-Zoom (PTZ) cameras, which is inspired by the chameleon vision of highly independent and symmetric, attention to both overall and local visual events. Because of the symmetry of two cameras and variability and controllability of camera parameters, compared with master-salve tracking system consisting of one static and one PTZ camera, the proposed system increases the surveillance coverage. Compared with system configuration using multiple static cameras and PTZ cameras, the proposed system decreases hardware cost. Compared with surveillance system composed of one omnidirectional and one PTZ camera, the proposed system is easy to fuse to exact more useful information. A master-slave control method is also designed using sphere coordinate model. Each camera can achieve cooperative master-slave tracking at arbitrary pan-tilt-zoom values. Therefore, visual attention of multi-resolution can be obtained. Quantitative results in the outdoor scene show good performance of the proposed approach.
This paper presents a master-salve tracking framework for wide area video surveillance with two Pan-Tilt-Zoom (PTZ) cameras, which is inspired by the chameleon vision of highly independent and symmetric, attention to both overall and local visual events. Because of the symmetry of two cameras and variability and controllability of camera parameters, compared with master-salve tracking system consisting of one static and one PTZ camera, the proposed system increases the surveillance coverage. Compared with system configuration using multiple static cameras and PTZ cameras, the proposed system decreases hardware cost. Compared with surveillance system composed of one omnidirectional and one PTZ camera, the proposed system is easy to fuse to exact more useful information. A master-slave control method is also designed using sphere coordinate model. Each camera can achieve cooperative master-slave tracking at arbitrary pan-tilt-zoom values. Therefore, visual attention of multi-resolution can be obtained. Quantitative results in the outdoor scene show good performance of the proposed approach.
2013, 35(4): 784-790.
doi: 10.3724/SP.J.1146.2012.01078
Abstract:
Considering the position and direction of trajectories of moving objects, a trajectory classification algorithm is proposed based on improved Hausdorff distance and Longest Common SubSequence (LCSS) to improve the trajectories classification. In this algorithm, the position similarity between trajectories is measured by the modified Hausdorff distances. And then the direction of the trajectories is distinguished by the modified LCSS distances. Comparing with other trajectory classification algorithms, the proposed algorithm compromises the merits of both Hausdorff distance and LCSS in trajectory classification and enhances the trajectory classification accuracy. Furthermore, to reduce the computational complexity of the similarity measure, a method of modified isometric transformation algorithm and an LCSS fast algorithm are realized. Experimental results show that the clustering accuracy of the proposed algorithm is greatly improved and the clustering accuracy rate can achieve 96%. Meanwhile, the computational cost is greatly reduced by the modified isometric transformation algorithm and the LCSS fast algorithm, and the magnitude of the declines can reach to 80%. The proposed algorithm can satisfy the system requirements of higher precision, real time and robustness.
Considering the position and direction of trajectories of moving objects, a trajectory classification algorithm is proposed based on improved Hausdorff distance and Longest Common SubSequence (LCSS) to improve the trajectories classification. In this algorithm, the position similarity between trajectories is measured by the modified Hausdorff distances. And then the direction of the trajectories is distinguished by the modified LCSS distances. Comparing with other trajectory classification algorithms, the proposed algorithm compromises the merits of both Hausdorff distance and LCSS in trajectory classification and enhances the trajectory classification accuracy. Furthermore, to reduce the computational complexity of the similarity measure, a method of modified isometric transformation algorithm and an LCSS fast algorithm are realized. Experimental results show that the clustering accuracy of the proposed algorithm is greatly improved and the clustering accuracy rate can achieve 96%. Meanwhile, the computational cost is greatly reduced by the modified isometric transformation algorithm and the LCSS fast algorithm, and the magnitude of the declines can reach to 80%. The proposed algorithm can satisfy the system requirements of higher precision, real time and robustness.
2013, 35(4): 791-796.
doi: 10.3724/SP.J.1146.2012.00759
Abstract:
Active contour models are successfully and widely used in image segmentation. However, they always get local minima which make wrong segmentation results. In this paper, based on previous work named background removed model, a convex energy which is obtained by approximating the Heaviside function in the previous nonconvex energy is proposed. By minimizing it, the evolution equation is given. Experimental results show that the proposed method is accurate, fast and antinoise. Moreover, it is not sensitive to the location of the initial curve.
Active contour models are successfully and widely used in image segmentation. However, they always get local minima which make wrong segmentation results. In this paper, based on previous work named background removed model, a convex energy which is obtained by approximating the Heaviside function in the previous nonconvex energy is proposed. By minimizing it, the evolution equation is given. Experimental results show that the proposed method is accurate, fast and antinoise. Moreover, it is not sensitive to the location of the initial curve.
2013, 35(4): 797-804.
doi: 10.3724/SP.J.1146.2012.00914
Abstract:
To balance high embedding capacity and imperceptibility of reversible watermarking algorithm for digital images, a novel Reversible Watermarking Algorithm based on Block Adaptive Compressed Sensing (BACS-RWA) is proposed. The host image is divided into blocks and the types of these blocks are determined with the statistical relationship between the surrounding image blocks and the target block. The capacity parameters are adaptively selected to do block compressed sensing and the watermarking is embedded with integer transformation. In order to improve embedding capacity, the smooth and normal blocks of compressed sensing host image are used to embed watermarking. Complex blocks are not processed to insure image quality and imperceptibility. Reconstruction algorithm of block compressed sensing and reversible integer transformation are used to reconstruct the host image accurately. Simulation of this algorithm is performed on different texture images and compared with similar algorithms. Experimental results show that the optimal embedding capacity can reach up to 1.87 bpp when Plane is used as host image. The introduction of block adaptive compressed sensing theory leads to better comprehensive performance. It can not only improve embedding capacity, but also reduce effectively the influence of embedding data on the quality of the host image.
To balance high embedding capacity and imperceptibility of reversible watermarking algorithm for digital images, a novel Reversible Watermarking Algorithm based on Block Adaptive Compressed Sensing (BACS-RWA) is proposed. The host image is divided into blocks and the types of these blocks are determined with the statistical relationship between the surrounding image blocks and the target block. The capacity parameters are adaptively selected to do block compressed sensing and the watermarking is embedded with integer transformation. In order to improve embedding capacity, the smooth and normal blocks of compressed sensing host image are used to embed watermarking. Complex blocks are not processed to insure image quality and imperceptibility. Reconstruction algorithm of block compressed sensing and reversible integer transformation are used to reconstruct the host image accurately. Simulation of this algorithm is performed on different texture images and compared with similar algorithms. Experimental results show that the optimal embedding capacity can reach up to 1.87 bpp when Plane is used as host image. The introduction of block adaptive compressed sensing theory leads to better comprehensive performance. It can not only improve embedding capacity, but also reduce effectively the influence of embedding data on the quality of the host image.
2013, 35(4): 805-811.
doi: 10.3724/SP.J.1146.2012.00949
Abstract:
It is always important issue to extract features that are most effective for preserving the distribution architecture in pattern recognition community. Kernel based methods are assumed to extract nonlinear features. However, it is very sensitive to the selection of its mapping function and parameters. This paper proposes a feature extraction algorithm based on multi-layer auto-encoder, which consists of two phases of unsupervised pretraining and supervised fine-tuning based on marginal Fisher rule. Generative pretraining and regularization methods within fine-tuning phase are adopted to avoid overfitting of models training. The validity of algorithm is proved within the result of classification experiments in several datasets.
It is always important issue to extract features that are most effective for preserving the distribution architecture in pattern recognition community. Kernel based methods are assumed to extract nonlinear features. However, it is very sensitive to the selection of its mapping function and parameters. This paper proposes a feature extraction algorithm based on multi-layer auto-encoder, which consists of two phases of unsupervised pretraining and supervised fine-tuning based on marginal Fisher rule. Generative pretraining and regularization methods within fine-tuning phase are adopted to avoid overfitting of models training. The validity of algorithm is proved within the result of classification experiments in several datasets.
2013, 35(4): 812-819.
doi: 10.3724/SP.J.1146.2012.00947
Abstract:
Demosaicking is important for the quality of digital images in resource-constrained single chip devices. This paper presents an improved dictionary learning-based color demosaicking algorithm. Firstly, an initial interpolation is applied to the,channel by Local Directional Interpolation (LDI) and fused by analysis the joint distribution of the gradient. Gaussian Mixture Model (GMM)-based clustering is used to classify dictionary image into different classes. The Principal Component Analysis (PCA) is performed on these classes to choose the principal components for the dictionary construction. And then, dictionary learning is applied to obtain the interpolatedG^ and the lostR^ and B^ are interpolated by the help of the reconstructed G^, accordingly. Since, R^, G^andB^ of the given pixels are better represented, the whole image can be reconstructed accurately. Taking McMaster color image dataset as dictionary, standard image and image from DALSA CMOS camera are used for effect evaluation of the demosaicking algorithm. Experimental results prove that the proposed algorithm outperforms some state-of-the-art demosaicking methods both in PSNR measure and visual quality.
Demosaicking is important for the quality of digital images in resource-constrained single chip devices. This paper presents an improved dictionary learning-based color demosaicking algorithm. Firstly, an initial interpolation is applied to the,channel by Local Directional Interpolation (LDI) and fused by analysis the joint distribution of the gradient. Gaussian Mixture Model (GMM)-based clustering is used to classify dictionary image into different classes. The Principal Component Analysis (PCA) is performed on these classes to choose the principal components for the dictionary construction. And then, dictionary learning is applied to obtain the interpolatedG^ and the lostR^ and B^ are interpolated by the help of the reconstructed G^, accordingly. Since, R^, G^andB^ of the given pixels are better represented, the whole image can be reconstructed accurately. Taking McMaster color image dataset as dictionary, standard image and image from DALSA CMOS camera are used for effect evaluation of the demosaicking algorithm. Experimental results prove that the proposed algorithm outperforms some state-of-the-art demosaicking methods both in PSNR measure and visual quality.
2013, 35(4): 820-825.
doi: 10.3724/SP.J.1146.2012.00589
Abstract:
By improving the Spatial-Temporal Total Variation (ST-TV) method, a video image reconstruction approach based on the Spatial Temporal Weighted Total Variation (ST-WTV) is proposed in this paper. By introducing ST-WTV as a regular term, a new model is got for video image sequences reconstruction. An algorithm based on split Bregman iterative method is given in this paper. Finally, the simulated and real data experimental results show that the proposed spatially ST-WTV video restoration algorithm not only efficiently reduces the artifacts produced with a TV model in fat regions of the image, but also preserves the edge information, getting more nature and detail-preserving image sequences.
By improving the Spatial-Temporal Total Variation (ST-TV) method, a video image reconstruction approach based on the Spatial Temporal Weighted Total Variation (ST-WTV) is proposed in this paper. By introducing ST-WTV as a regular term, a new model is got for video image sequences reconstruction. An algorithm based on split Bregman iterative method is given in this paper. Finally, the simulated and real data experimental results show that the proposed spatially ST-WTV video restoration algorithm not only efficiently reduces the artifacts produced with a TV model in fat regions of the image, but also preserves the edge information, getting more nature and detail-preserving image sequences.
2013, 35(4): 826-831.
doi: 10.3724/SP.J.1146.2012.00921
Abstract:
Fast Alternating Direction Method of Multipliers (FADMM) is proposed to solve the l0-regularisation issue, which is a problem of signal compression and reconstruction for Compressed Sensing (CS). The first step of FADMM is to express the l0-regularisation issue of the sparse coefficient as a constrained optimization issue by using variable splitting technology. Then, by introducing the function of multipliers, the two variables are alternatively minimized by Gauss-Seidel method. And the two variables are updated once again to speed up the convergence rate, and then, the variable of multipliers is updated. Finally, the original signal is reconstructed by the orthogonal inverse transform. FADMM is better than other state-of-the-art algorithms on reconstructing image. And the experimental simulations demonstrate that the FADMM algorithm has a higher Peak Signal to Noise Ratio (PSNR) and a faster convergence rate.
Fast Alternating Direction Method of Multipliers (FADMM) is proposed to solve the l0-regularisation issue, which is a problem of signal compression and reconstruction for Compressed Sensing (CS). The first step of FADMM is to express the l0-regularisation issue of the sparse coefficient as a constrained optimization issue by using variable splitting technology. Then, by introducing the function of multipliers, the two variables are alternatively minimized by Gauss-Seidel method. And the two variables are updated once again to speed up the convergence rate, and then, the variable of multipliers is updated. Finally, the original signal is reconstructed by the orthogonal inverse transform. FADMM is better than other state-of-the-art algorithms on reconstructing image. And the experimental simulations demonstrate that the FADMM algorithm has a higher Peak Signal to Noise Ratio (PSNR) and a faster convergence rate.
2013, 35(4): 832-837.
doi: 10.3724/SP.J.1146.2012.01047
Abstract:
Precise synchronization can be achieved by fitting phase discrimination curve exploiting Least Squares (LS). However, this method not only has very high requirement of coarse synchronization, but also can not counteract frequency- offset effectively. It is analyzed and proved that the measured value of LS is between zero and real value as coarse synchronization error is between the earlier and later chip. Accordingly a precise synchronization method based on iterative least squares is proposed and a modulus of split correlation strategy is also introduced in this paper to counteract frequency-offset. Theoretical analysis shows that this method can outstandingly eliminate influence of noise as well as frequency-offset with short segments. Besides its strong anti frequency-offset and noise capability, the proposed method still has very high measurement precision as the error of coarse synchronization covers the earlier and later half chip according to simulation results.
Precise synchronization can be achieved by fitting phase discrimination curve exploiting Least Squares (LS). However, this method not only has very high requirement of coarse synchronization, but also can not counteract frequency- offset effectively. It is analyzed and proved that the measured value of LS is between zero and real value as coarse synchronization error is between the earlier and later chip. Accordingly a precise synchronization method based on iterative least squares is proposed and a modulus of split correlation strategy is also introduced in this paper to counteract frequency-offset. Theoretical analysis shows that this method can outstandingly eliminate influence of noise as well as frequency-offset with short segments. Besides its strong anti frequency-offset and noise capability, the proposed method still has very high measurement precision as the error of coarse synchronization covers the earlier and later half chip according to simulation results.
2013, 35(4): 838-843.
doi: 10.3724/SP.J.1146.2012.00912
Abstract:
This paper discusses choice for scaling parameter of the unscented transformation. By analyzing and comparing some scaling parameter selection methods, the scaling parameter is selected as an optimization objective. Differential Evolution (DE) algorithm is applied to the Unscented Kalman Filter (UKF), the optimized scaling parameter leads to the minimum error at each time interval. An adaptive UKF based on DE is proposed. The experiments show that the accuracy of UKF is significantly improved by the adaptive strategy which not only to avoid random divergence with the constant parameter but also suitable for any form of UKF without the constraints of the number of parameters.
This paper discusses choice for scaling parameter of the unscented transformation. By analyzing and comparing some scaling parameter selection methods, the scaling parameter is selected as an optimization objective. Differential Evolution (DE) algorithm is applied to the Unscented Kalman Filter (UKF), the optimized scaling parameter leads to the minimum error at each time interval. An adaptive UKF based on DE is proposed. The experiments show that the accuracy of UKF is significantly improved by the adaptive strategy which not only to avoid random divergence with the constant parameter but also suitable for any form of UKF without the constraints of the number of parameters.
2013, 35(4): 844-851.
doi: 10.3724/SP.J.1146.2012.01008
Abstract:
The analysis of the target radiated noise model is made as a precondition under the background of fluctuating sound field firstly. A broad and narrow band integrated coherent detection algorithm based on phase fluctuation and differential alignment is proposed for the issue of target azimuth estimation and narrow-band line-spectrum detection. The background noise energy disturbances can be restrained by using the differences of temporal correlated radius and phase fluctuation uniformity for the periodic spectrum signals and broadband noise; The ability to identify different DOA targets can also be improved by combining phase fluctuation aligned gain of the output signal in focused beam domain and spatial coherent gain of the signal processing in array domain. It is proved that the proposed algorithm can significantly enhance the detection SNR gain of the target line spectral component and relative energy spectrum level of the coherent targets beam azimuth by analyzing the simulation and sea experiment results. It has better application prospects in the detection and identification areas of anti-complex channel.
The analysis of the target radiated noise model is made as a precondition under the background of fluctuating sound field firstly. A broad and narrow band integrated coherent detection algorithm based on phase fluctuation and differential alignment is proposed for the issue of target azimuth estimation and narrow-band line-spectrum detection. The background noise energy disturbances can be restrained by using the differences of temporal correlated radius and phase fluctuation uniformity for the periodic spectrum signals and broadband noise; The ability to identify different DOA targets can also be improved by combining phase fluctuation aligned gain of the output signal in focused beam domain and spatial coherent gain of the signal processing in array domain. It is proved that the proposed algorithm can significantly enhance the detection SNR gain of the target line spectral component and relative energy spectrum level of the coherent targets beam azimuth by analyzing the simulation and sea experiment results. It has better application prospects in the detection and identification areas of anti-complex channel.
2013, 35(4): 852-858.
doi: 10.3724/SP.J.1146.2012.00607
Abstract:
Because of the special track of circular SAR, the imaging algorithms of SAR in linear track can not be used in processing the circular SAR data directly. However, the response functions of the circular SAR systems can remain the consistency by changing the same aspect angle both of the target and the starting position of the radar. Because of this characteristic, circular SAR can be processed in the azimuth frequency domain by using the fast imaging algorithm of SAR in linear track for reference. In this paper, the exact signal spectrum expressions of circular SAR are derived. By using the reasonable approximations of the exact signal spectrum expressions, the fast imaging algorithms are presented. The simulations show the good performance of the proposed algorithms.
Because of the special track of circular SAR, the imaging algorithms of SAR in linear track can not be used in processing the circular SAR data directly. However, the response functions of the circular SAR systems can remain the consistency by changing the same aspect angle both of the target and the starting position of the radar. Because of this characteristic, circular SAR can be processed in the azimuth frequency domain by using the fast imaging algorithm of SAR in linear track for reference. In this paper, the exact signal spectrum expressions of circular SAR are derived. By using the reasonable approximations of the exact signal spectrum expressions, the fast imaging algorithms are presented. The simulations show the good performance of the proposed algorithms.
2013, 35(4): 859-864.
doi: 10.3724/SP.J.1146.2012.01121
Abstract:
Owning to the fact that the high speed moving target moves across several range cells during the long coherent observation period, joint Direction Of Departures (DODs) and Direction Of Arrivals (DOAs) estimation with the echo data located at the single rang cell has considerable error that affect the cross-positioning for the high speed moving targets in bistatic MIMO radar. A method for estimating the parameters of high speed moving targets is presented. The average of the sample covariance matrixes estimated from the pulse compression outputs at different range cells is exploited to obtain the high precision estimation of the covariance matrix. Then, DODs and DOAs of targets can be estimated by using the traditional supper-resolution algorithm. The simulation results indicate that the angle estimation accuracy of the proposed method for the high speed moving targets, which is independent of the number of range cells that target moves across, is better than that of the existence approaches, and is close to that of the existence approaches with no range migration occurrence.
Owning to the fact that the high speed moving target moves across several range cells during the long coherent observation period, joint Direction Of Departures (DODs) and Direction Of Arrivals (DOAs) estimation with the echo data located at the single rang cell has considerable error that affect the cross-positioning for the high speed moving targets in bistatic MIMO radar. A method for estimating the parameters of high speed moving targets is presented. The average of the sample covariance matrixes estimated from the pulse compression outputs at different range cells is exploited to obtain the high precision estimation of the covariance matrix. Then, DODs and DOAs of targets can be estimated by using the traditional supper-resolution algorithm. The simulation results indicate that the angle estimation accuracy of the proposed method for the high speed moving targets, which is independent of the number of range cells that target moves across, is better than that of the existence approaches, and is close to that of the existence approaches with no range migration occurrence.
2013, 35(4): 865-870.
doi: 10.3724/SP.J.1146.2012.01573
Abstract:
The common predictions of angular glint are usually based on the condition of far-field hypothesis. However, computation of near field angular glint is very meaningful?both in academic studies and engineering applications due to its huge impact on tracking error. This paper presents an analytic prediction of near field angular glint computing based on Graphic Electromagnetic Computing (GRECO), which is both in real time and on omission of the process of scattering center extraction. The condition that targets cannot be covered by radar beam is firstly considered, which makes the prediction of tracking error more complete. The correctness is validated by comparison between the simulated results and the theoretical ones of several models.
The common predictions of angular glint are usually based on the condition of far-field hypothesis. However, computation of near field angular glint is very meaningful?both in academic studies and engineering applications due to its huge impact on tracking error. This paper presents an analytic prediction of near field angular glint computing based on Graphic Electromagnetic Computing (GRECO), which is both in real time and on omission of the process of scattering center extraction. The condition that targets cannot be covered by radar beam is firstly considered, which makes the prediction of tracking error more complete. The correctness is validated by comparison between the simulated results and the theoretical ones of several models.
2013, 35(4): 871-876.
doi: 10.3724/SP.J.1146.2012.01011
Abstract:
Using passive radar with Orthogonal Frequency Division Multiplexing (OFDM) waveform for target track and location is recently a research spot at home and abroad. This paper first introduces the signal model for passive radar with OFDM waveform and key techniques during the signal processing. On that basis, how the estimation error of carrier frequency offset influences the detection performance of this radar is mainly investigated through simulation from different aspects, including the influences on target parameter estimation after match filtering, clutter rejection performance and bit error rate during reference signal reconstruction. The simulation results show that different temporal clutter rejection methods require different estimation accuracy of carrier frequency offset. The reconstruction bit error rate has a great impact on the temporal clutter rejection performance while having a little effect on match filtering. Finally, the analysis results are verified based on the measurement data.
Using passive radar with Orthogonal Frequency Division Multiplexing (OFDM) waveform for target track and location is recently a research spot at home and abroad. This paper first introduces the signal model for passive radar with OFDM waveform and key techniques during the signal processing. On that basis, how the estimation error of carrier frequency offset influences the detection performance of this radar is mainly investigated through simulation from different aspects, including the influences on target parameter estimation after match filtering, clutter rejection performance and bit error rate during reference signal reconstruction. The simulation results show that different temporal clutter rejection methods require different estimation accuracy of carrier frequency offset. The reconstruction bit error rate has a great impact on the temporal clutter rejection performance while having a little effect on match filtering. Finally, the analysis results are verified based on the measurement data.
2013, 35(4): 877-881.
doi: 10.3724/SP.J.1146.2012.00797
Abstract:
The information of targets Direction Of Arriving (DOA) is very important for target location. But in passive radar, the weak target echoes are usually embedded in the background of strong direct signal, multipath and noise. So it is very difficult to estimate the DOA of target in passive radar, especially when there are multiple targets. In this paper, a method based on compressed sensing is proposed to perform high resolution DOA estimation in passive radar. In order to remove the strong direct signal and multipath and improve the SNR of targets, first the temporal interference cancellation and range-Doppler 2D correlation are utilized. Finally, the signal of targets delay-Doppler bin is reconstructed according to DOA using reconstruction algorithm of compressed sensing. Simulation results show that the proposed method can perform the super-resolution DOA estimation in passive radar.
The information of targets Direction Of Arriving (DOA) is very important for target location. But in passive radar, the weak target echoes are usually embedded in the background of strong direct signal, multipath and noise. So it is very difficult to estimate the DOA of target in passive radar, especially when there are multiple targets. In this paper, a method based on compressed sensing is proposed to perform high resolution DOA estimation in passive radar. In order to remove the strong direct signal and multipath and improve the SNR of targets, first the temporal interference cancellation and range-Doppler 2D correlation are utilized. Finally, the signal of targets delay-Doppler bin is reconstructed according to DOA using reconstruction algorithm of compressed sensing. Simulation results show that the proposed method can perform the super-resolution DOA estimation in passive radar.
2013, 35(4): 882-887.
doi: 10.3724/SP.J.1146.2012.00927
Abstract:
Based on the fluctuation analysis, a novel approach for target detection in sea clutter is proposed. The self-affinity and scaling behaviors of sea clutter is analyzed by using the mean fluctuation. The q order normalized slope of fluctuation curve, as the characteristic parameter, is suggested to describe the fractal property of the target and sea clutter. The tests on the real data show that the target could be clearly distinguished from the sea clutter background with the proposed approach.
Based on the fluctuation analysis, a novel approach for target detection in sea clutter is proposed. The self-affinity and scaling behaviors of sea clutter is analyzed by using the mean fluctuation. The q order normalized slope of fluctuation curve, as the characteristic parameter, is suggested to describe the fractal property of the target and sea clutter. The tests on the real data show that the target could be clearly distinguished from the sea clutter background with the proposed approach.
2013, 35(4): 888-893.
doi: 10.3724/SP.J.1146.2012.01462
Abstract:
Range-dependent clutter suppression is one of the most important problem in the airborne bistatic radar Space-Time Adaptive Processing (STAP). To overcome the problem, a novel technique using overlapped subarray alternate transmitting is developed. By virtue of the alternate transmitting, the clutter ridges of all range bins of airborne bistatic sidelooking radar are located in the same plane of a three-dimensional space which is constructed by transmitting spatial frequency, receiving frequency and Doppler frequency. Due to the range-independence of the clutter plane, it can be well suppressed by STAP. Then range-dependent clutter is efficiently suppressed. Numerical examples are given to demonstrate the effectiveness of the presented method.
Range-dependent clutter suppression is one of the most important problem in the airborne bistatic radar Space-Time Adaptive Processing (STAP). To overcome the problem, a novel technique using overlapped subarray alternate transmitting is developed. By virtue of the alternate transmitting, the clutter ridges of all range bins of airborne bistatic sidelooking radar are located in the same plane of a three-dimensional space which is constructed by transmitting spatial frequency, receiving frequency and Doppler frequency. Due to the range-independence of the clutter plane, it can be well suppressed by STAP. Then range-dependent clutter is efficiently suppressed. Numerical examples are given to demonstrate the effectiveness of the presented method.
2013, 35(4): 894-900.
doi: 10.3724/SP.J.1146.2012.01026
Abstract:
Classification of moving vehicles within short dwell time is a promising way to the introduction of the target identification function to battlefield surveillance radar system. In this paper, radar returned echoes of moving wheeled and tracked vehicle are analyzed using micro-Doppler effect. According to the distinction between the micro-Doppler signals of these two kinds of vehicles, a wavelet transform based classification method is proposed. In this method, the influence induced by the change of main bulk velocity is alleviated by using multirate signal processing and the distinctions between wheeled and tracked vehicles are well depicted due to the separation of the bulk motion and micro motion components. Experiment results based on the measured data show the proposed method simultaneously achieves good classification performance and robustness to the change of the bulk velocity.
Classification of moving vehicles within short dwell time is a promising way to the introduction of the target identification function to battlefield surveillance radar system. In this paper, radar returned echoes of moving wheeled and tracked vehicle are analyzed using micro-Doppler effect. According to the distinction between the micro-Doppler signals of these two kinds of vehicles, a wavelet transform based classification method is proposed. In this method, the influence induced by the change of main bulk velocity is alleviated by using multirate signal processing and the distinctions between wheeled and tracked vehicles are well depicted due to the separation of the bulk motion and micro motion components. Experiment results based on the measured data show the proposed method simultaneously achieves good classification performance and robustness to the change of the bulk velocity.
2013, 35(4): 901-907.
doi: 10.3724/SP.J.1146.2012.00883
Abstract:
For the requirement of making full use of the limited resource in real application, this paper proposes a power allocation scheme for 3D target tracking in multistatic radar systems. Firstly, the Bayesian Cramer Rao Lower Bound (BCRLB) for 3D target tracking mean square error is derived and utilized as a criterion for the power allocation strategy. Then, the resulting nonlinear convex problem is solved by gradient projection algorithm. Finally, simulation results suggest that the tracking accuracy of the whole system is enormously improved by using the proposed algorithm compared with the results achieved with equal power allocation.
For the requirement of making full use of the limited resource in real application, this paper proposes a power allocation scheme for 3D target tracking in multistatic radar systems. Firstly, the Bayesian Cramer Rao Lower Bound (BCRLB) for 3D target tracking mean square error is derived and utilized as a criterion for the power allocation strategy. Then, the resulting nonlinear convex problem is solved by gradient projection algorithm. Finally, simulation results suggest that the tracking accuracy of the whole system is enormously improved by using the proposed algorithm compared with the results achieved with equal power allocation.
2013, 35(4): 908-914.
doi: 10.3724/SP.J.1146.2012.01058
Abstract:
Presently, the radar altimeter look-up correction tables are usually computed with an approximate exponential flat sea surface response function, and the effects of this approximation are neglected. In this paper, the errors of this approximation method are analyzed, and a new method of using the accurate flat sea surface response function is derived to compute the look-up correction tables. This new method can greatly decrease the errors when antenna mispointing angle is large. The new method is applied to HY-2 altimeter, and obtains a more accurate result.
Presently, the radar altimeter look-up correction tables are usually computed with an approximate exponential flat sea surface response function, and the effects of this approximation are neglected. In this paper, the errors of this approximation method are analyzed, and a new method of using the accurate flat sea surface response function is derived to compute the look-up correction tables. This new method can greatly decrease the errors when antenna mispointing angle is large. The new method is applied to HY-2 altimeter, and obtains a more accurate result.
2013, 35(4): 915-920.
doi: 10.3724/SP.J.1146.2012.01200
Abstract:
The systemic bias of space-based sensor hinders accurate threat identification and target location of coming targets. The correction of this systematic bias has unique difficulties including unable to in-situ commissioning and systemic bias periodically change as the satellite undergoes a cyclical heating and cooling due to its orbit. Combining the satellite altitude determination system, this paper firstly obtains the star vector measurement from electro-optical sensor and monitors the deviation of these measurements from expected value in navigation star table. Then, on the basis of systemic bias, this paper derives bias model and design the Bias Corrected Shift Rayleigh Filter (BCSRF). Simulation results show that the proposed filter can achieve in-situ calibration, and yields significant improvements in tracking ballistic targets compared with classical intersection and Unscented Kalman Filter (UKF) methods.
The systemic bias of space-based sensor hinders accurate threat identification and target location of coming targets. The correction of this systematic bias has unique difficulties including unable to in-situ commissioning and systemic bias periodically change as the satellite undergoes a cyclical heating and cooling due to its orbit. Combining the satellite altitude determination system, this paper firstly obtains the star vector measurement from electro-optical sensor and monitors the deviation of these measurements from expected value in navigation star table. Then, on the basis of systemic bias, this paper derives bias model and design the Bias Corrected Shift Rayleigh Filter (BCSRF). Simulation results show that the proposed filter can achieve in-situ calibration, and yields significant improvements in tracking ballistic targets compared with classical intersection and Unscented Kalman Filter (UKF) methods.
2013, 35(4): 921-926.
doi: 10.3724/SP.J.1146.2012.01065
Abstract:
This paper propses a filtering method for GPS/DR (Global Positioning System/Dead-Reckoning) integrated navigation system with unknown parameters. This method firstly structures a self-organizing state space model, and then estimates the state vector by using Monte Carlo filtering method for this new system model. Because particle filter is easy to make a search of the unknown parameters into a subset of the initial sampling for the self-organization model an artificial fish swarm-partical filter algorithm is put forward. The algorithm not only can estimate the system state, but also can make the sampling distribution of the unknown parameters move to the true parameter distribution. Ultimately, the true value of the unknown parameters are identified. The simuliation results show the effectiveness of the proposed method.
This paper propses a filtering method for GPS/DR (Global Positioning System/Dead-Reckoning) integrated navigation system with unknown parameters. This method firstly structures a self-organizing state space model, and then estimates the state vector by using Monte Carlo filtering method for this new system model. Because particle filter is easy to make a search of the unknown parameters into a subset of the initial sampling for the self-organization model an artificial fish swarm-partical filter algorithm is put forward. The algorithm not only can estimate the system state, but also can make the sampling distribution of the unknown parameters move to the true parameter distribution. Ultimately, the true value of the unknown parameters are identified. The simuliation results show the effectiveness of the proposed method.
Fast Image Reconstruction for Non-uniform Sampling of Thinned Array of Synthesis Aperture Radiometer
2013, 35(4): 927-932.
doi: 10.3724/SP.J.1146.2012.00688
Abstract:
Some thinned array, such as circle thinned array, is used by synthetic aperture interferometric radiometer to realize time-shared sampling. The thinned array may take non-uniform sampling of spatial frequencies due to antenna array structure optimization, the limitation of affined area and the physical size of antenna elements as well. In some traditional image reconstruction methods, non-uniform spatial frequency samplings are usually inserted to uniform Cartesian coordinate. And the spatial frequency densities are compensated. However, as the spatial sampling intensity is assumed to be quite smooth, these methods still bring errors and blurring in spite of different interpolation functions. Moreover, iterative methods that adopted straight Fourier transform are time consuming as they applied to non-uniform spatial frequency samplings. In this paper, another fast interactive image reconstruction method is introduced. The kernel of this algorithm is min-max formulation. The specific procedure of this method is as follows: (1)initiating the image; (2)taking non-uniform fast Fourier transform; (3)operating iterative conjugate gradient matched algorithm. The numerical simulating experiments show that, as spatial frequency samplings density is not smooth, the image can be still reestablished fast and accurately.
Some thinned array, such as circle thinned array, is used by synthetic aperture interferometric radiometer to realize time-shared sampling. The thinned array may take non-uniform sampling of spatial frequencies due to antenna array structure optimization, the limitation of affined area and the physical size of antenna elements as well. In some traditional image reconstruction methods, non-uniform spatial frequency samplings are usually inserted to uniform Cartesian coordinate. And the spatial frequency densities are compensated. However, as the spatial sampling intensity is assumed to be quite smooth, these methods still bring errors and blurring in spite of different interpolation functions. Moreover, iterative methods that adopted straight Fourier transform are time consuming as they applied to non-uniform spatial frequency samplings. In this paper, another fast interactive image reconstruction method is introduced. The kernel of this algorithm is min-max formulation. The specific procedure of this method is as follows: (1)initiating the image; (2)taking non-uniform fast Fourier transform; (3)operating iterative conjugate gradient matched algorithm. The numerical simulating experiments show that, as spatial frequency samplings density is not smooth, the image can be still reestablished fast and accurately.
2013, 35(4): 933-938.
doi: 10.3724/SP.J.1146.2012.01029
Abstract:
To guarantee the QoS of multi-hop wireless networks, it is necessary to analyse the delay bound of the system. Based on stochastic network calculus, this paper models traffic flowing scenarios of multi-hop wireless networks utilizing three flow operators and proposes a general analysis method for the delay performance. The method simplifies the analysis of complex traffic flowing scenarios and is capable of deriving the delay bound of different traffic flowing scenarios. Applying the method to a traffic flowing scenario, the method can accurately estimate the probability distribution of the actual delay. Furthermore, the analytical bound is very close to the simulation results and is obviously superior to the deterministic upper delay bound.
To guarantee the QoS of multi-hop wireless networks, it is necessary to analyse the delay bound of the system. Based on stochastic network calculus, this paper models traffic flowing scenarios of multi-hop wireless networks utilizing three flow operators and proposes a general analysis method for the delay performance. The method simplifies the analysis of complex traffic flowing scenarios and is capable of deriving the delay bound of different traffic flowing scenarios. Applying the method to a traffic flowing scenario, the method can accurately estimate the probability distribution of the actual delay. Furthermore, the analytical bound is very close to the simulation results and is obviously superior to the deterministic upper delay bound.
2013, 35(4): 939-945.
doi: 10.3724/SP.J.1146.2012.01088
Abstract:
Dynamic Spectrum Allocation (DSA) in heterogeneous wireless networks is investigated for the downlink communication of cellular Base Stations (BSs). A new Interference Control (IC) model for the case of log-normal shadowing is proposed based on Coverage Probability (CP). Such IC model considers the users distribution and transforms the SIR (Signal to Interference Ratio) requirement of users with different services into the coverage performance of BSs. Based on the IC model, DSA is formulated as a combinatorial optimization problem with nonlinear constraints. An algorithm of greedy CP-DSA is designed for the DSA issue. Simulation results indicate that CP-DSA can increase the network utility, control effectively the interference among BSs and satisfy the SIR requirement of users.
Dynamic Spectrum Allocation (DSA) in heterogeneous wireless networks is investigated for the downlink communication of cellular Base Stations (BSs). A new Interference Control (IC) model for the case of log-normal shadowing is proposed based on Coverage Probability (CP). Such IC model considers the users distribution and transforms the SIR (Signal to Interference Ratio) requirement of users with different services into the coverage performance of BSs. Based on the IC model, DSA is formulated as a combinatorial optimization problem with nonlinear constraints. An algorithm of greedy CP-DSA is designed for the DSA issue. Simulation results indicate that CP-DSA can increase the network utility, control effectively the interference among BSs and satisfy the SIR requirement of users.
2013, 35(4): 946-951.
doi: 10.3724/SP.J.1146.2012.01110
Abstract:
The demand of broadband access to mobile Internet has experienced rapid growth in recent years, which brings both great opportunities and challenges to the mobile operators. It is the operators focus of concern to find optimized pricing mechanism to guide the mobile user,s behavior and improve their revenue. This paper, with the analysis of mobile user behavior and the influence of price on it, not only proves that only by using time-variant pricing can the operators maximize their revenue, but also gives some methods to guide mobile user behavior. Considering the users resistance to the frequent pricing changes, this paper has also examined the feasibility of half time-variant pricing, and shows how to get the loss rate of network revenue. Research results show that half time-variant pricing can effectively mitigate the operators revenue loss while maintaining appropriate user acceptability.
The demand of broadband access to mobile Internet has experienced rapid growth in recent years, which brings both great opportunities and challenges to the mobile operators. It is the operators focus of concern to find optimized pricing mechanism to guide the mobile user,s behavior and improve their revenue. This paper, with the analysis of mobile user behavior and the influence of price on it, not only proves that only by using time-variant pricing can the operators maximize their revenue, but also gives some methods to guide mobile user behavior. Considering the users resistance to the frequent pricing changes, this paper has also examined the feasibility of half time-variant pricing, and shows how to get the loss rate of network revenue. Research results show that half time-variant pricing can effectively mitigate the operators revenue loss while maintaining appropriate user acceptability.
2013, 35(4): 952-957.
doi: 10.3724/SP.J.1146.2012.00983
Abstract:
With the rapid growth of network traffic, applications of the multi-path transmission are becoming more popular. The multi-path routing algorithms are the base to implement multi-path transmission. In this paper, a new Multi-path routing algorithm based on the A* algorithm called MA* algorithm is proposed which combining the path similarity objective with the heuristic method. A new evaluation function construction method is devised for the MA* algorithm, which is proved to be feasible through theoretical analysis. The simulation experiments are carried out to compare the MA* algorithm with other multi-path routing algorithms applying network topology generation tool. The results show that the MA* algorithm can acquire multiple paths with reasonable path cost and path similarity at quite low search times. In addition, the balance of path cost and path similarity can be achieved through reasonable adjustment of the punishing parameter. The MA* algorithm can be applied to wider multi-path transmission environments as a more common multi-path routing algorithm.
With the rapid growth of network traffic, applications of the multi-path transmission are becoming more popular. The multi-path routing algorithms are the base to implement multi-path transmission. In this paper, a new Multi-path routing algorithm based on the A* algorithm called MA* algorithm is proposed which combining the path similarity objective with the heuristic method. A new evaluation function construction method is devised for the MA* algorithm, which is proved to be feasible through theoretical analysis. The simulation experiments are carried out to compare the MA* algorithm with other multi-path routing algorithms applying network topology generation tool. The results show that the MA* algorithm can acquire multiple paths with reasonable path cost and path similarity at quite low search times. In addition, the balance of path cost and path similarity can be achieved through reasonable adjustment of the punishing parameter. The MA* algorithm can be applied to wider multi-path transmission environments as a more common multi-path routing algorithm.
2013, 35(4): 958-964.
doi: 10.3724/SP.J.1146.2012.01040
Abstract:
Considering at the concept drift issue of machine learning identification, a novel algorithm called traffic classification based on Host Connection Graph (HCG) is proposed. Considering {IP Address, Port} as the unique user identifier, HCG constructs a host connection graph and innovates the concept of user similarity. Based on the theory of graph mining, social community is abstracted from communications among hosts by partitioning the graph into mutually intersectant behavior clusters. In order to reach traffic classification, HCG not only conceives a definition called User Behavior Mode (UBM) to analyse the implicit traffic characteristics, but also maps application labels to every host behavior by employing UBM and Port. Finally, simulations are conducted based on the real network trace. Results demonstrate that HCG can circumvent the concept shift problem and ameliorate gracefully computational complication without sacrificing accuracy.
Considering at the concept drift issue of machine learning identification, a novel algorithm called traffic classification based on Host Connection Graph (HCG) is proposed. Considering {IP Address, Port} as the unique user identifier, HCG constructs a host connection graph and innovates the concept of user similarity. Based on the theory of graph mining, social community is abstracted from communications among hosts by partitioning the graph into mutually intersectant behavior clusters. In order to reach traffic classification, HCG not only conceives a definition called User Behavior Mode (UBM) to analyse the implicit traffic characteristics, but also maps application labels to every host behavior by employing UBM and Port. Finally, simulations are conducted based on the real network trace. Results demonstrate that HCG can circumvent the concept shift problem and ameliorate gracefully computational complication without sacrificing accuracy.
2013, 35(4): 965-969.
doi: 10.3724/SP.J.1146.2012.01128
Abstract:
With the rapid development of heterogeneous wireless multimedia sensor networks, how to satisfy all the requirement of data source and provide QoS for it becomes a hot research issue. Differentiated Queuing Service (DQS) is introduced into wireless multimedia sensor network. Based on the algorithm, scalar node can inform sink in time and real time traffic can be forwarded rapidly through setting packet lifetime. The result of simulation show that the sink can gather enough event scalar packets and real time packets delay is satisfied. The effectiveness of the method is verified.
With the rapid development of heterogeneous wireless multimedia sensor networks, how to satisfy all the requirement of data source and provide QoS for it becomes a hot research issue. Differentiated Queuing Service (DQS) is introduced into wireless multimedia sensor network. Based on the algorithm, scalar node can inform sink in time and real time traffic can be forwarded rapidly through setting packet lifetime. The result of simulation show that the sink can gather enough event scalar packets and real time packets delay is satisfied. The effectiveness of the method is verified.
2013, 35(4): 970-976.
doi: 10.3724/SP.J.1146.2012.00940
Abstract:
Privacy preservation in wireless sensor networks has attracted more and more attentions. Computing MAX/MIN query result in wireless sensor networks while preserving data privacy is a challenge. This paper proposes a Zero-One (Z-O) encoding based Privacy-Preserving MAX/MIN query protocol in two-tiered wireless sensor networks (ZOPPM). In ZOPPM, sensor nodes in the query range firstly convert their sensory data into encoded data, by using Z-O encoding and hashed message authentication code mechanism, and send the encoded data to the corresponding storage node, and encrypt sensory data and send the ciphertext to it in the demand. According to the numerical comparison theory of Z-O encoding method, every storage node generates a local MAX/MIN sensor node in its own query cell, without sensory data in plaintext. Then, the storage node constructs a local query result when receiving the encrypted data from a sensor node, and sends it to the sink node. Finally, the sink node calculates the MAX/MIN query result after receiving the local query result from all storage nodes. The theoretical analysis and experimental results show that the ZOPPM protocol can ensure the privacy of sensory data and the query result, and it costs less energy consumption than other similar method.
Privacy preservation in wireless sensor networks has attracted more and more attentions. Computing MAX/MIN query result in wireless sensor networks while preserving data privacy is a challenge. This paper proposes a Zero-One (Z-O) encoding based Privacy-Preserving MAX/MIN query protocol in two-tiered wireless sensor networks (ZOPPM). In ZOPPM, sensor nodes in the query range firstly convert their sensory data into encoded data, by using Z-O encoding and hashed message authentication code mechanism, and send the encoded data to the corresponding storage node, and encrypt sensory data and send the ciphertext to it in the demand. According to the numerical comparison theory of Z-O encoding method, every storage node generates a local MAX/MIN sensor node in its own query cell, without sensory data in plaintext. Then, the storage node constructs a local query result when receiving the encrypted data from a sensor node, and sends it to the sink node. Finally, the sink node calculates the MAX/MIN query result after receiving the local query result from all storage nodes. The theoretical analysis and experimental results show that the ZOPPM protocol can ensure the privacy of sensory data and the query result, and it costs less energy consumption than other similar method.
2013, 35(4): 977-981.
doi: 10.3724/SP.J.1146.2012.01118
Abstract:
Community mining in dynamic networks can help to obtain the whole network characteristics and the trend of network development. As dynamic networks usually consist of many consecutive static networks, traditional methods of identifying network communities will lead to significant variations between communities close in time and high time complexity. Although the general incremental methods (e.g. Incremental algorithm for Community identification (IC) and Increment and Density based Community detection Method (IDCM)) can reduce the time complexity at a certain extent, but they need to manually set the judgment parameter, and fail to identify large networks in acceptable time. In this paper, an algorithm of integrating Random Walk and Increment correction Vertexes (RWIV) is proposed to identify the dynamic network structure. RWIV algorithm first deals with increment correlative vertexes with random walk, and then adjusts the residual vertexes by analyzing their community affinity. The simulation results and analysis show that RWIV avoid manually selecting the parameter of IC or IDCM, which affect the accuracy of community mining, the cumulative error. RWIV can fit the situation of community structure sharp changes. The performance of RWIV is super than that of IC and IDCM methods.
Community mining in dynamic networks can help to obtain the whole network characteristics and the trend of network development. As dynamic networks usually consist of many consecutive static networks, traditional methods of identifying network communities will lead to significant variations between communities close in time and high time complexity. Although the general incremental methods (e.g. Incremental algorithm for Community identification (IC) and Increment and Density based Community detection Method (IDCM)) can reduce the time complexity at a certain extent, but they need to manually set the judgment parameter, and fail to identify large networks in acceptable time. In this paper, an algorithm of integrating Random Walk and Increment correction Vertexes (RWIV) is proposed to identify the dynamic network structure. RWIV algorithm first deals with increment correlative vertexes with random walk, and then adjusts the residual vertexes by analyzing their community affinity. The simulation results and analysis show that RWIV avoid manually selecting the parameter of IC or IDCM, which affect the accuracy of community mining, the cumulative error. RWIV can fit the situation of community structure sharp changes. The performance of RWIV is super than that of IC and IDCM methods.
2013, 35(4): 982-987.
doi: 10.3724/SP.J.1146.2012.01282
Abstract:
Load balancing is a fundamental problem for cloud computing, multi-processing load balancing session persistence in multi-core environment have drawn more attention and have become a focus. For the issue, based on the idea of Hash Linux kernel network data packets passing, a lock-free multi-processing load balancing architecture is proposed, which avoids the use of locks, and can quickly change the existing single-processing load balancing procedure for multi-processing architecture without structural changes. The theory analysis and experimental results show that the proposed architecture is able to improve the overall performance of load balancing system in multi-core environment. Compared with the traditional shared memory architecture, the proposed is able to get better performance and has stronger applicability.
Load balancing is a fundamental problem for cloud computing, multi-processing load balancing session persistence in multi-core environment have drawn more attention and have become a focus. For the issue, based on the idea of Hash Linux kernel network data packets passing, a lock-free multi-processing load balancing architecture is proposed, which avoids the use of locks, and can quickly change the existing single-processing load balancing procedure for multi-processing architecture without structural changes. The theory analysis and experimental results show that the proposed architecture is able to improve the overall performance of load balancing system in multi-core environment. Compared with the traditional shared memory architecture, the proposed is able to get better performance and has stronger applicability.
2013, 35(4): 988-993.
doi: 10.3724/SP.J.1146.2012.01102
Abstract:
Fully homomorphic encryption has important application in cloud computing. However, the existing fully homomorphic encryption schemes share a common flaw that they all use public keys of large scales. And this flaw may cause inefficiency of these schemes in the key and identity management. To solve this problem, a certificateless fully homomorphic encryption scheme is presented based on Learning With Errors (LWE) problem. The scheme builds the connection between the users identity and its public key with the trapdoor one-way function with preimage sampling so that the certificates are no longer necessary. The private keys are chosen by the users without key escrow. In the random oracle model, the security of the scheme strictly reduces to hardness of decisional LWE problem.
Fully homomorphic encryption has important application in cloud computing. However, the existing fully homomorphic encryption schemes share a common flaw that they all use public keys of large scales. And this flaw may cause inefficiency of these schemes in the key and identity management. To solve this problem, a certificateless fully homomorphic encryption scheme is presented based on Learning With Errors (LWE) problem. The scheme builds the connection between the users identity and its public key with the trapdoor one-way function with preimage sampling so that the certificates are no longer necessary. The private keys are chosen by the users without key escrow. In the random oracle model, the security of the scheme strictly reduces to hardness of decisional LWE problem.
2013, 35(4): 994-1000.
doi: 10.3724/SP.J.1146.2012.01030
Abstract:
To solve the problem of soft error caused by Single Event Upset (SEU) in Static Random Access Memory (SRAM)-based Field Programmable Gate Arrays (FPGAs), the impact of routing resources by Single Bit Upset (SBU) and Multiple Bit Upset (MBU) is analyzed. A new method of soft-error-mitigation physical design approach is presented. In the approach, the error probability of routing resources is introduced for evaluation soft error. Combined with error propagation probability, system failure rate is calculated for driving placement and routing. The experimental results show that the system failure rate decreases about 18% using proposed method. This method can also effectively mitigate effect of multiple bit upset.
To solve the problem of soft error caused by Single Event Upset (SEU) in Static Random Access Memory (SRAM)-based Field Programmable Gate Arrays (FPGAs), the impact of routing resources by Single Bit Upset (SBU) and Multiple Bit Upset (MBU) is analyzed. A new method of soft-error-mitigation physical design approach is presented. In the approach, the error probability of routing resources is introduced for evaluation soft error. Combined with error propagation probability, system failure rate is calculated for driving placement and routing. The experimental results show that the system failure rate decreases about 18% using proposed method. This method can also effectively mitigate effect of multiple bit upset.
2013, 35(4): 1001-1006.
doi: 10.3724/SP.J.1146.2012.00650
Abstract:
Periods with high branch misprediction rate tend to be uneven and concentrated during execution of programs. To address this problem, a new branch prediction strategy is proposed, which based on dynamic polarity transformation. This approach monitors original branch misprediction rate whose polarity has not been transformed, and detects the periods with original branch misprediction rate higher than a threshold. These periods are called as peaks of misprediction. The polarity of original prediction results will be transformed to make the final prediction during peaks of misprediction. As a result, the final branch misprediction rate whose polarity has been transformed will always be lower than the threshold during execution of programs. The prediction method can be divided into three categories according to the monitoring mechanism, which are global monitor, set monitor and per-address monitor. The experimental results show that this methodology gives better prediction accuracy than Gshare and Bi-Mode prediction schemes for the same cost.
Periods with high branch misprediction rate tend to be uneven and concentrated during execution of programs. To address this problem, a new branch prediction strategy is proposed, which based on dynamic polarity transformation. This approach monitors original branch misprediction rate whose polarity has not been transformed, and detects the periods with original branch misprediction rate higher than a threshold. These periods are called as peaks of misprediction. The polarity of original prediction results will be transformed to make the final prediction during peaks of misprediction. As a result, the final branch misprediction rate whose polarity has been transformed will always be lower than the threshold during execution of programs. The prediction method can be divided into three categories according to the monitoring mechanism, which are global monitor, set monitor and per-address monitor. The experimental results show that this methodology gives better prediction accuracy than Gshare and Bi-Mode prediction schemes for the same cost.
2013, 35(4): 1007-1011.
doi: 10.3724/SP.J.1146.2012.01090
Abstract:
A modeling and analysis scheme based on Markov chain for Random Access MAC (RA-MAC) protocol of cognitive radio networks are proposed. A three-state Markov chain model of channel is presented, while the problem of dynamic characteristic of channel is dealt with by introducing the channel constraint rate and the available channel number of stability. A channel negotiation model is designed based on discrete two-dimensional Markov chain. An algorithm to evaluate the saturation throughput of RA-MAC is presented, considering constraints of capacity of the control channel. The simulation results demonstrate the effectiveness of the proposals, and the relationship between protocol performance and parameters is analyzed.
A modeling and analysis scheme based on Markov chain for Random Access MAC (RA-MAC) protocol of cognitive radio networks are proposed. A three-state Markov chain model of channel is presented, while the problem of dynamic characteristic of channel is dealt with by introducing the channel constraint rate and the available channel number of stability. A channel negotiation model is designed based on discrete two-dimensional Markov chain. An algorithm to evaluate the saturation throughput of RA-MAC is presented, considering constraints of capacity of the control channel. The simulation results demonstrate the effectiveness of the proposals, and the relationship between protocol performance and parameters is analyzed.
2013, 35(4): 1012-1016.
doi: 10.3724/SP.J.1146.2012.01106
Abstract:
In order to solve the problem that the covert timing channel works unstable in the noisy network, a method of designing robust covert channel is proposed. The method uses the interval time of network packets to transfer information, the sender codes covert information using hash spreading spectrum, and the receiver forecasts the channel noise and eliminates it. In order to solve the contradictory relationship between transmission rate and robustness, the strategy of maximizing robustness under fixed transmission rate is proposed. The experimental environment of this covert channel is constructed and several experiments are conducted. The results show that the ability to resist noise is increased about by 20%, compared with other methods on the same problem.
In order to solve the problem that the covert timing channel works unstable in the noisy network, a method of designing robust covert channel is proposed. The method uses the interval time of network packets to transfer information, the sender codes covert information using hash spreading spectrum, and the receiver forecasts the channel noise and eliminates it. In order to solve the contradictory relationship between transmission rate and robustness, the strategy of maximizing robustness under fixed transmission rate is proposed. The experimental environment of this covert channel is constructed and several experiments are conducted. The results show that the ability to resist noise is increased about by 20%, compared with other methods on the same problem.
