Email alert
2013 Vol. 35, No. 3
Display Method:
2013, 35(3): 509-515.
doi: 10.3724/SP.J.1146.2012.00992
Abstract:
The seriously expended clutter Doppler and finite sample supports are the problem of high velocity radar platform, and the performance of conventional Space Time Adaptive Processing (STAP) will degrade due to covariance matrix error or inaccurate target constraint. A research on clutter Degree of Freedom (DoF) of forward-looking radar is discussed. The eigenvalue spectrum of clutter covariance matrix is relative with aperture bandwidth product, time-width bandwidth product together with the correlation of spatial and temporal frequency. Robust STAP method based on space-time multi-beam transforming is proposed. First, the DoF of clutter is reduced through space-time multi-beam transforming. Then, the extended Doppler constrain method is used to improve the performance and robustness of the processer. The numerical simulations show the effectiveness of the proposed method. Even when the training data contains the target signal, the proposed robust method can maintain up to 5 dB Signal-to-Clutter-plus-Noise Ratio (SCNR) improvement compared with the conventional methods.
The seriously expended clutter Doppler and finite sample supports are the problem of high velocity radar platform, and the performance of conventional Space Time Adaptive Processing (STAP) will degrade due to covariance matrix error or inaccurate target constraint. A research on clutter Degree of Freedom (DoF) of forward-looking radar is discussed. The eigenvalue spectrum of clutter covariance matrix is relative with aperture bandwidth product, time-width bandwidth product together with the correlation of spatial and temporal frequency. Robust STAP method based on space-time multi-beam transforming is proposed. First, the DoF of clutter is reduced through space-time multi-beam transforming. Then, the extended Doppler constrain method is used to improve the performance and robustness of the processer. The numerical simulations show the effectiveness of the proposed method. Even when the training data contains the target signal, the proposed robust method can maintain up to 5 dB Signal-to-Clutter-plus-Noise Ratio (SCNR) improvement compared with the conventional methods.
2013, 35(3): 516-523.
doi: 10.3724/SP.J.1146.2012.00444
Abstract:
Generally the problem about Doppler Blind Zone (DBZ) occurs in airborne early warning radar. In the process of target tracking, the DBZ can cause track termination and new track initiation easily. Considering the problem, a target tracking method based on Multiple-Hypothesis Motion Models (MHMM) is proposed. The method produces multiple hypothesis motion models according to target state restriction set by DBZ. Once new measurements fall into association gate formed according to a certain model, the track association is successful. The simulation results show that on the condition of different DBZ ranges, high track association ratio can be acquired in this method for targets with different maneuverability. This algorithm can improve performance of continuous target tracking.
Generally the problem about Doppler Blind Zone (DBZ) occurs in airborne early warning radar. In the process of target tracking, the DBZ can cause track termination and new track initiation easily. Considering the problem, a target tracking method based on Multiple-Hypothesis Motion Models (MHMM) is proposed. The method produces multiple hypothesis motion models according to target state restriction set by DBZ. Once new measurements fall into association gate formed according to a certain model, the track association is successful. The simulation results show that on the condition of different DBZ ranges, high track association ratio can be acquired in this method for targets with different maneuverability. This algorithm can improve performance of continuous target tracking.
2013, 35(3): 524-531.
doi: 10.3724/SP.J.1146.2012.00844
Abstract:
In order to set the detection threshold in real time according to the demanded false alarm probability for Bayesian track-before-detect under Neyman-Pearson criterion, this paper derives the closed-form solution of the detection threshold in white complex Gaussian noise. For the Bayesian track-before-detect, this paper starts from the likelihood ratio testing form, derives the relationship between the false alarm probability and the detection threshold in detail, and obtains the closed-form solution of the detection threshold in white complex Gaussian noise. The simulation results show that the detection threshold can be ascertained in real time for Bayesian track-before-detect according to the false alarm probability in demand using the presented approach.
In order to set the detection threshold in real time according to the demanded false alarm probability for Bayesian track-before-detect under Neyman-Pearson criterion, this paper derives the closed-form solution of the detection threshold in white complex Gaussian noise. For the Bayesian track-before-detect, this paper starts from the likelihood ratio testing form, derives the relationship between the false alarm probability and the detection threshold in detail, and obtains the closed-form solution of the detection threshold in white complex Gaussian noise. The simulation results show that the detection threshold can be ascertained in real time for Bayesian track-before-detect according to the false alarm probability in demand using the presented approach.
2013, 35(3): 532-536.
doi: 10.3724/SP.J.1146.2012.00857
Abstract:
A method of designing constant modulus waveforms with ultra-low range sidelobe in specified intervals is presented. This kind of waveform can be used in radar, sonar, and communication systems with transmit adaptivity to suppress range sidelobe masking and multipath interferences. The 0-1 weighted integrated sidelobe level is used to construct the objective function, and the waveform design issue is modeled as a non-constrained optimization issue. According to the characteristics of the objective function, an initial waveform design algorithm based on power spectrum approximation is proposed. The analytical forms of objective function gradient and the Hessian matrix are derived and the subspace trust region algorithm is used, thus the computation burden of solving this optimization issue is reduced. Finally, the numerical simulations demonstrate that the proposed method can suppress successive and discrete intervals of range sidelobes effectively.
A method of designing constant modulus waveforms with ultra-low range sidelobe in specified intervals is presented. This kind of waveform can be used in radar, sonar, and communication systems with transmit adaptivity to suppress range sidelobe masking and multipath interferences. The 0-1 weighted integrated sidelobe level is used to construct the objective function, and the waveform design issue is modeled as a non-constrained optimization issue. According to the characteristics of the objective function, an initial waveform design algorithm based on power spectrum approximation is proposed. The analytical forms of objective function gradient and the Hessian matrix are derived and the subspace trust region algorithm is used, thus the computation burden of solving this optimization issue is reduced. Finally, the numerical simulations demonstrate that the proposed method can suppress successive and discrete intervals of range sidelobes effectively.
2013, 35(3): 537-544.
doi: 10.3724/SP.J.1146.2012.00978
Abstract:
Feature parameters estimation for ballistic targets is the basis of target recognition. Because of the difficulty to estimate the precession and structure parameters of a cone-cylinder target jointly without prior knowledge of parameters, a new method for the cone-cylinder targets precession and structure parameters estimation is proposed based on High Range Resolution Profile sequences (HRRPs). The HRRPs of the rotary symmetrical cone-cylinder target are simulated and analyzed in four attitude angle regions respectively based on static electromagnetic scattering data. The trace of each scatter in the HRRPs is researched in the common radar viewing aspect, and the relationship between the extreme value of three scattering centers relative positions in the HRRPs and the targets parameters is built, then the precession and structure parameters are estimated simultaneously. Finally, the estimation results of feature parameters show that the proposed method is effective and adaptive.
Feature parameters estimation for ballistic targets is the basis of target recognition. Because of the difficulty to estimate the precession and structure parameters of a cone-cylinder target jointly without prior knowledge of parameters, a new method for the cone-cylinder targets precession and structure parameters estimation is proposed based on High Range Resolution Profile sequences (HRRPs). The HRRPs of the rotary symmetrical cone-cylinder target are simulated and analyzed in four attitude angle regions respectively based on static electromagnetic scattering data. The trace of each scatter in the HRRPs is researched in the common radar viewing aspect, and the relationship between the extreme value of three scattering centers relative positions in the HRRPs and the targets parameters is built, then the precession and structure parameters are estimated simultaneously. Finally, the estimation results of feature parameters show that the proposed method is effective and adaptive.
2013, 35(3): 545-551.
doi: 10.3724/SP.J.1146.2012.00944
Abstract:
In the frequency-stepped chirp signal, the frequency spectrum overlap between the neighboring sub pulses would lead to grating lobes in the result of range compression. The weaker points may be masked or distorted by the grating lobes of strong points. In this paper, a grating lobe suppression algorithm is proposed based on apodization filtering. By weighting the spectrum of sub pulse signals with different functions, the apodization could be done through the results and the grating lobes could be eliminated. Numerical simulations and real data experiment confirm the effectiveness of the proposed algorithm.
In the frequency-stepped chirp signal, the frequency spectrum overlap between the neighboring sub pulses would lead to grating lobes in the result of range compression. The weaker points may be masked or distorted by the grating lobes of strong points. In this paper, a grating lobe suppression algorithm is proposed based on apodization filtering. By weighting the spectrum of sub pulse signals with different functions, the apodization could be done through the results and the grating lobes could be eliminated. Numerical simulations and real data experiment confirm the effectiveness of the proposed algorithm.
2013, 35(3): 552-558.
doi: 10.3724/SP.J.1146.2012.01016
Abstract:
Side-looking three-Dimensional (3D) imaging of three-aperture sparse array Synthetic Aperture Radar (SAR) based on Compressed Sensing (CS) is investigated in this paper. Using the sparse array structure in cross- track direction formed by three aperture antennas, the elevation resolution can be obtained and 3D imaging is achieved. However, the conventional Fourier transform based 3D imaging approach has a low resolution in elevation direction and brings the image quality problem because of the low number of acquisitions and irregular space sampling. CS theory is introduced to improve the resolution in the elevation direction. Experiment results on simulation and real data validate the effectiveness of the proposed method compared with the conventional imaging technique.
Side-looking three-Dimensional (3D) imaging of three-aperture sparse array Synthetic Aperture Radar (SAR) based on Compressed Sensing (CS) is investigated in this paper. Using the sparse array structure in cross- track direction formed by three aperture antennas, the elevation resolution can be obtained and 3D imaging is achieved. However, the conventional Fourier transform based 3D imaging approach has a low resolution in elevation direction and brings the image quality problem because of the low number of acquisitions and irregular space sampling. CS theory is introduced to improve the resolution in the elevation direction. Experiment results on simulation and real data validate the effectiveness of the proposed method compared with the conventional imaging technique.
2013, 35(3): 559-567.
doi: 10.3724/SP.J.1146.2012.00850
Abstract:
InSAR motion compensation is a key step to obtain high-precision product. The effect of MOtion COmpensation (MOCO) errors on the precision of airborne dual-antenna InSAR is analysed. First, the MOCO residual errors are deduced on condition that the referenced DEM (Digital Elevation Model) is accurate or not, respectively. Next, the effects of MOCO errors on the interferometric phase and plane location are analyzed without measurement errors considered. Then the effects resulted from trajectory measurement errors and baseline measurement errors are presented. Finally, the accuracy of the theoretical deduction is verified with simulation and real data. The research provides theoretical bases for the system design and signal processing of airborne dual-antenna InSAR.
InSAR motion compensation is a key step to obtain high-precision product. The effect of MOtion COmpensation (MOCO) errors on the precision of airborne dual-antenna InSAR is analysed. First, the MOCO residual errors are deduced on condition that the referenced DEM (Digital Elevation Model) is accurate or not, respectively. Next, the effects of MOCO errors on the interferometric phase and plane location are analyzed without measurement errors considered. Then the effects resulted from trajectory measurement errors and baseline measurement errors are presented. Finally, the accuracy of the theoretical deduction is verified with simulation and real data. The research provides theoretical bases for the system design and signal processing of airborne dual-antenna InSAR.
2013, 35(3): 568-574.
doi: 10.3724/SP.J.1146.2012.00891
Abstract:
In conventional radar system, the resolution is constrained by Nyquist sampling rate. A large amount of data is created under the high-resolution requirement. Compressive Sensing (CS) relieves the demand of A/D converter and the capacity of memories. Under the framework of CS, a set of bases, which is incomplete but is based on the targets' features, is given out in this paper. A method is proposed for reconstruction that is compatible with the bases. The sparseness of the issue is not necessary for the proposed approach. And the method has very good performance on dealing with linear targets, especially when the lengths of the targets are very long. Furthermore, it can also resolve the multi-target issue. The simulation results verify the efficiency of the proposed algorithm.
In conventional radar system, the resolution is constrained by Nyquist sampling rate. A large amount of data is created under the high-resolution requirement. Compressive Sensing (CS) relieves the demand of A/D converter and the capacity of memories. Under the framework of CS, a set of bases, which is incomplete but is based on the targets' features, is given out in this paper. A method is proposed for reconstruction that is compatible with the bases. The sparseness of the issue is not necessary for the proposed approach. And the method has very good performance on dealing with linear targets, especially when the lengths of the targets are very long. Furthermore, it can also resolve the multi-target issue. The simulation results verify the efficiency of the proposed algorithm.
2013, 35(3): 575-580.
doi: 10.3724/SP.J.1146.2012.00939
Abstract:
Digital Terrestrial Multimedia Broadcasting (DTMB) is announced as the Chinese own independent intellectual proprietary digital terrestrial TV standard. A passive radar system is introduced for target detection based on DTMB. The principle, key techniques, experimental equipment and preliminary results are given. The technical feasibility of using DTMB signal for target detection is proved by field experiment.
Digital Terrestrial Multimedia Broadcasting (DTMB) is announced as the Chinese own independent intellectual proprietary digital terrestrial TV standard. A passive radar system is introduced for target detection based on DTMB. The principle, key techniques, experimental equipment and preliminary results are given. The technical feasibility of using DTMB signal for target detection is proved by field experiment.
2013, 35(3): 581-588.
doi: 10.3724/SP.J.1146.2012.00903
Abstract:
For passive radar, the computation cost of traditional 2-dimentional coherent integration (ambiguity function) is high, which makes it hard to perform in real-time processing. And range migration may occur during the integration time, leading to a Signal-to-Noise ratio (SNR) decrease. The migration may cause a detection ability loss. A migration compensation algorithm is proposed with keystone transform based on a real-time integration method in this article to realize real-time processing of 2-dimational integration. The real-time integration includes: 1-dimentional signal is divided into segments, match filtering is applied to each segment, and FFT is applied across segments. An improved segment method is introduced for application demand. Based on this method, an effective method to rectify range migration is proposed using keystone transform in this article, enhancing the detection ability of weak targets. The algorithm can improve computation efficiency with less energy loss, and is easier to perform in real-time then other algorithms. So its a real-time and efficient target detection method. In this article the real-time processing ability of integration and migration compensation is analyzed. Experiments based on simulated signal and real-life signal show that the algorithm can realize real-time processing with effective processing gain.
For passive radar, the computation cost of traditional 2-dimentional coherent integration (ambiguity function) is high, which makes it hard to perform in real-time processing. And range migration may occur during the integration time, leading to a Signal-to-Noise ratio (SNR) decrease. The migration may cause a detection ability loss. A migration compensation algorithm is proposed with keystone transform based on a real-time integration method in this article to realize real-time processing of 2-dimational integration. The real-time integration includes: 1-dimentional signal is divided into segments, match filtering is applied to each segment, and FFT is applied across segments. An improved segment method is introduced for application demand. Based on this method, an effective method to rectify range migration is proposed using keystone transform in this article, enhancing the detection ability of weak targets. The algorithm can improve computation efficiency with less energy loss, and is easier to perform in real-time then other algorithms. So its a real-time and efficient target detection method. In this article the real-time processing ability of integration and migration compensation is analyzed. Experiments based on simulated signal and real-life signal show that the algorithm can realize real-time processing with effective processing gain.
2013, 35(3): 589-594.
doi: 10.3724/SP.J.1146.2012.00900
Abstract:
By calculating the cross ambiguity function of the reference signal and the echo signal based on the passive coherent location technologies, passive radar detects the distance and speed of the target. However, it is difficult to meet real-time requirements when calculated the cross ambiguity function directly. To solve this problem, this paper proposes a method that uses the rectangular window as the first stage filter of a multi-stage decimation based on two types of fast calculation method: delay traverse and Doppler frequency traverse, which are based on Chinese National Standard Digital Television Terrestrial Broadcasting (DTTB) signal. Compared with the direct calculation and the non-decimation calculation method, the amount of the calculation can be reduced more than 3 orders and 1 order separately. This paper further analyses the variance of the calculation amount of the two types of fast calculation method when time accumulation is changing and provides a reference for the choice of calculation methods in practical applications.
By calculating the cross ambiguity function of the reference signal and the echo signal based on the passive coherent location technologies, passive radar detects the distance and speed of the target. However, it is difficult to meet real-time requirements when calculated the cross ambiguity function directly. To solve this problem, this paper proposes a method that uses the rectangular window as the first stage filter of a multi-stage decimation based on two types of fast calculation method: delay traverse and Doppler frequency traverse, which are based on Chinese National Standard Digital Television Terrestrial Broadcasting (DTTB) signal. Compared with the direct calculation and the non-decimation calculation method, the amount of the calculation can be reduced more than 3 orders and 1 order separately. This paper further analyses the variance of the calculation amount of the two types of fast calculation method when time accumulation is changing and provides a reference for the choice of calculation methods in practical applications.
2013, 35(3): 595-600.
doi: 10.3724/SP.J.1146.2012.00630
Abstract:
The spectrum of the clutter is spatially-temporally coupled in the airborne MIMO radar. Based on the knowledge that the location of the ground clutter in angle-Doppler domain is mainly dependent on the velocity of the airplane and the radar parameters, a novel two dimensional pulse-to-pulse canceller is designed to suppress the clutter signals more efficiently. As a pre-filtering tool before the conventional Moving Target Indication (MTI) and the sub-optimal dimension-reduced Spatial-Temporal Adaptive Processing (STAP) algorithms, the method can enhance the performance of moving target detection. The experimental results demonstrate that the proposed pre-filtering approach can effectively suppress the clutter and can be flexibly constructed according to different array configurations. Moreover, the proposed method can gain a better performance for detecting the high speed moving target.
The spectrum of the clutter is spatially-temporally coupled in the airborne MIMO radar. Based on the knowledge that the location of the ground clutter in angle-Doppler domain is mainly dependent on the velocity of the airplane and the radar parameters, a novel two dimensional pulse-to-pulse canceller is designed to suppress the clutter signals more efficiently. As a pre-filtering tool before the conventional Moving Target Indication (MTI) and the sub-optimal dimension-reduced Spatial-Temporal Adaptive Processing (STAP) algorithms, the method can enhance the performance of moving target detection. The experimental results demonstrate that the proposed pre-filtering approach can effectively suppress the clutter and can be flexibly constructed according to different array configurations. Moreover, the proposed method can gain a better performance for detecting the high speed moving target.
2013, 35(3): 601-607.
doi: 10.3724/SP.J.1146.2012.01072
Abstract:
Focusing on the radar allocation for stealth targets detection and tracking issue in air-defense radar network, a novel collaborative detection and tracking algorithm that combines Binary Particle Swarm Optimization (BPSO) and particle filtering to cope with the radar allocation is proposed in this paper. In the proposed algorithm, Radar Allocation Schemes (RAS) are designed according to the characters of stealthy targets, and the particles distributed randomly are applied to obtain the detection probability of newborn targets. Then the tracking accuracy is measured by the Posterior Cramr-Rao Lower Bound (PCRLB) of the tracked targets. Moreover, the BPSO is selected to search the whole RAS, and the results of particle filtering of the selected tracking radars are fused. Simulation results show that the proposed method can not only quickly identify newborn targets, but also optimize the tracking performance of the existing targets, and improve the tracking accuracy of the whole radar network compared with traditional methods.
Focusing on the radar allocation for stealth targets detection and tracking issue in air-defense radar network, a novel collaborative detection and tracking algorithm that combines Binary Particle Swarm Optimization (BPSO) and particle filtering to cope with the radar allocation is proposed in this paper. In the proposed algorithm, Radar Allocation Schemes (RAS) are designed according to the characters of stealthy targets, and the particles distributed randomly are applied to obtain the detection probability of newborn targets. Then the tracking accuracy is measured by the Posterior Cramr-Rao Lower Bound (PCRLB) of the tracked targets. Moreover, the BPSO is selected to search the whole RAS, and the results of particle filtering of the selected tracking radars are fused. Simulation results show that the proposed method can not only quickly identify newborn targets, but also optimize the tracking performance of the existing targets, and improve the tracking accuracy of the whole radar network compared with traditional methods.
Chaotic Analog-to-information Conversion: Sparse Signal Reconstruction with Multiple Shooting Method
2013, 35(3): 608-613.
doi: 10.3724/SP.J.1146.2012.00905
Abstract:
Chaotic Compressive Sensing (CS) is a nonlinear compressive sensing theory which utilizes the randomness-like characteristic of chaos systems to measure sparse signals. This paper focuses on the chaotic compressive sensing for the acquisition and reconstruction of analog signals, i.e., Chaotic Analog-to-Information (ChaA2I) converter. ChaA2I generates the low-rate samples by sampling the output of chaotic system excited by the sparse signals, and implements the signal reconstruction by solving the sparsity-regularized nonlinear least squares problem. With the view on chaotic parameter estimation, a highly-efficient reconstruction algorithm (MS-IRNLS) is developed by combing the Multiple Shooting (MS) method with the Iteratively Reweighted Nonlinear Least-Squares (IRNLS) algorithm. With the Lorenz system as an example, the paper conducts extensive simulations for the reconstruction performance of MS-IRNLS algorithm. The simulations demonstrate the effectiveness of the proposed ChaA2I.
Chaotic Compressive Sensing (CS) is a nonlinear compressive sensing theory which utilizes the randomness-like characteristic of chaos systems to measure sparse signals. This paper focuses on the chaotic compressive sensing for the acquisition and reconstruction of analog signals, i.e., Chaotic Analog-to-Information (ChaA2I) converter. ChaA2I generates the low-rate samples by sampling the output of chaotic system excited by the sparse signals, and implements the signal reconstruction by solving the sparsity-regularized nonlinear least squares problem. With the view on chaotic parameter estimation, a highly-efficient reconstruction algorithm (MS-IRNLS) is developed by combing the Multiple Shooting (MS) method with the Iteratively Reweighted Nonlinear Least-Squares (IRNLS) algorithm. With the Lorenz system as an example, the paper conducts extensive simulations for the reconstruction performance of MS-IRNLS algorithm. The simulations demonstrate the effectiveness of the proposed ChaA2I.
2013, 35(3): 614-621.
doi: 10.3724/SP.J.1146.2012.00892
Abstract:
In order to enhance the global optimization capability and quorum sensing mechanism of Bacterial Foraging Optimization (BFO) algorithm, a novel Bacterial Foraging Optimization algorithm with Quantum Behavior (QBFO) is proposed. In this method, the bacteria individual is described in the quantum space and a potential well model is created. Using Monte Carlo method to achieve the reproduction of bacterial swarming, and which makes the population are able to search the whole space. In view of the defects of the fixed swim step in bacterial foraging algorithm, a dynamic indented control strategy is introduced in this paper, which ensures the convergence of algorithm and increases the possibility of exploring a global optimum. The experiment results on classic functions demonstrate the global convergence ability of the proposed method with better accuracy and more probability of finding global optimum.
In order to enhance the global optimization capability and quorum sensing mechanism of Bacterial Foraging Optimization (BFO) algorithm, a novel Bacterial Foraging Optimization algorithm with Quantum Behavior (QBFO) is proposed. In this method, the bacteria individual is described in the quantum space and a potential well model is created. Using Monte Carlo method to achieve the reproduction of bacterial swarming, and which makes the population are able to search the whole space. In view of the defects of the fixed swim step in bacterial foraging algorithm, a dynamic indented control strategy is introduced in this paper, which ensures the convergence of algorithm and increases the possibility of exploring a global optimum. The experiment results on classic functions demonstrate the global convergence ability of the proposed method with better accuracy and more probability of finding global optimum.
2013, 35(3): 622-626.
doi: 10.3724/SP.J.1146.2012.01218
Abstract:
This paper presents a novel multi-instance multi-label image classification method based on sparse coding and ensemble learning. First, a dictionary is learned based on all the instances in the training bags, and the sparse coding coefficient of each instance is calculated according to the dictionary; Second, a bag feature vector is computed based on all the sparse coding coefficients of the bag. Multi-instance multi-label issue is transformed into multi-label issue that can be solved by the multi-label algorithm. Ensemble learning is involved to enhance further the classifiers generalization. Experimental results on multi-instance multi-label image data show that the proposed method is superior to the state-of-art methods in terms of metrics.
This paper presents a novel multi-instance multi-label image classification method based on sparse coding and ensemble learning. First, a dictionary is learned based on all the instances in the training bags, and the sparse coding coefficient of each instance is calculated according to the dictionary; Second, a bag feature vector is computed based on all the sparse coding coefficients of the bag. Multi-instance multi-label issue is transformed into multi-label issue that can be solved by the multi-label algorithm. Ensemble learning is involved to enhance further the classifiers generalization. Experimental results on multi-instance multi-label image data show that the proposed method is superior to the state-of-art methods in terms of metrics.
2013, 35(3): 627-632.
doi: 10.3724/SP.J.1146.2012.00016
Abstract:
The Citation-kNN algorithm improves traditional kNN algorithm and can be applied to solve multi- instance learning issue. But its 0-1 decision strategy has some limitations. To overcome this issue, the locally-weighted Citation-kNN algorithm is presented in this paper. Considering distribution of the samples, the distance-based weighted method and the scatter-based weighted method are proposed. And their combinations are discussed. The method is applied to the standard database MUSK and the breast ultrasound image database. The results confirm that the method has higher accuracy comparing with that by using Citation-kNN algorithm.
The Citation-kNN algorithm improves traditional kNN algorithm and can be applied to solve multi- instance learning issue. But its 0-1 decision strategy has some limitations. To overcome this issue, the locally-weighted Citation-kNN algorithm is presented in this paper. Considering distribution of the samples, the distance-based weighted method and the scatter-based weighted method are proposed. And their combinations are discussed. The method is applied to the standard database MUSK and the breast ultrasound image database. The results confirm that the method has higher accuracy comparing with that by using Citation-kNN algorithm.
2013, 35(3): 633-638.
doi: 10.3724/SP.J.1146.2012.00793
Abstract:
As a dimensionality reduction algorithm, Local Fisher Discriminant Analysis (LFDA) is faced with two problems: (1) how to select the favorable neighborhood size which may have effect on the optimal projection direction and (2) the neglect of neighborhood relationships between samples of different classes. In order to overcome the drawback of LFDA, a novel dimensionality reduction algorithm called neighborhood graph embedding based Local Adaptive Discriminant Projection (LADP) is proposed in this paper. First, LADP adaptively estimates within-class and between-class neighborhood set according to samples, distribution and similarity. Then local weighted matrices are defined depending on the neighborhood size. Ultimately optimal embedding subspace is gained by maximizing local between-class scatter and minimizing local within-class scatter. LADP can preserve both local information and discriminant information. The experimental results of the toy example and real-word data validate the effectiveness of the proposed algorithm.
As a dimensionality reduction algorithm, Local Fisher Discriminant Analysis (LFDA) is faced with two problems: (1) how to select the favorable neighborhood size which may have effect on the optimal projection direction and (2) the neglect of neighborhood relationships between samples of different classes. In order to overcome the drawback of LFDA, a novel dimensionality reduction algorithm called neighborhood graph embedding based Local Adaptive Discriminant Projection (LADP) is proposed in this paper. First, LADP adaptively estimates within-class and between-class neighborhood set according to samples, distribution and similarity. Then local weighted matrices are defined depending on the neighborhood size. Ultimately optimal embedding subspace is gained by maximizing local between-class scatter and minimizing local within-class scatter. LADP can preserve both local information and discriminant information. The experimental results of the toy example and real-word data validate the effectiveness of the proposed algorithm.
2013, 35(3): 639-644.
doi: 10.3724/SP.J.1146.2012.00866
Abstract:
The surgery assisted robotic tool helps the surgeon to cancel the relative motion between the beating heart and robotic tool, keeping the heart beating during the surgery, which will lessen post surgery complications for patients. Due to the highly irregular and non-stationary nature of heart motion, the robot is hard to track the beating heart motion. To solve this problem, a characteristic analysis of 3D heart motion data through Bi-spectral tool is used to demonstrate the nonlinearity of coupling between respiration and heartbeat in heart motion. Then an nonlinear Second order Volterra Series (SVS) based fast least square prediction algorithm is proposed to provide the future reference to the controller. The nonlinear model would accurately describe the heart motion and the fast least square algorithm would satisfy the real time needs. The comparative experiment results indicate that the proposed adaptive nonlinear heart motion prediction algorithm outperforms the former algorithms in the term of prediction accuracy. The relative motion cancellation ability of the robot is enhanced and prediction error is largely reduced.
The surgery assisted robotic tool helps the surgeon to cancel the relative motion between the beating heart and robotic tool, keeping the heart beating during the surgery, which will lessen post surgery complications for patients. Due to the highly irregular and non-stationary nature of heart motion, the robot is hard to track the beating heart motion. To solve this problem, a characteristic analysis of 3D heart motion data through Bi-spectral tool is used to demonstrate the nonlinearity of coupling between respiration and heartbeat in heart motion. Then an nonlinear Second order Volterra Series (SVS) based fast least square prediction algorithm is proposed to provide the future reference to the controller. The nonlinear model would accurately describe the heart motion and the fast least square algorithm would satisfy the real time needs. The comparative experiment results indicate that the proposed adaptive nonlinear heart motion prediction algorithm outperforms the former algorithms in the term of prediction accuracy. The relative motion cancellation ability of the robot is enhanced and prediction error is largely reduced.
2013, 35(3): 645-651.
doi: 10.3724/SP.J.1146.2012.00673
Abstract:
Considering the complexity and the accuracy, an improved affinity propagation clustering algorithm Semi-supervised Affinity Propagation clustering algorithm based on Stratified Combination (SAP-SC) is proposed. In order to make the operation simplified and easily-implemented, the proposed algorithm introduces a stratified clustering method which equally partitions the integrative clustering process into several smaller blocks. Focusing on the hard clustering data, every layer employs semi-supervised learning to conceive pair-wise constraints and maps each sub-cluster with the corresponding label. Also, assembled boosting method is utilized to weight together all layered results to improve the clustering performance. Finally, theoretical analysis and experimental results show that the algorithm can achieve both higher accuracy and better computational performance.
Considering the complexity and the accuracy, an improved affinity propagation clustering algorithm Semi-supervised Affinity Propagation clustering algorithm based on Stratified Combination (SAP-SC) is proposed. In order to make the operation simplified and easily-implemented, the proposed algorithm introduces a stratified clustering method which equally partitions the integrative clustering process into several smaller blocks. Focusing on the hard clustering data, every layer employs semi-supervised learning to conceive pair-wise constraints and maps each sub-cluster with the corresponding label. Also, assembled boosting method is utilized to weight together all layered results to improve the clustering performance. Finally, theoretical analysis and experimental results show that the algorithm can achieve both higher accuracy and better computational performance.
2013, 35(3): 652-657.
doi: 10.3724/SP.J.1146.2012.00831
Abstract:
In non-destructive testing, ultrasonic echo is often an overlapping multi-echoes signal with noise. A time- frequency estimation algorithm for ultrasonic echo signal baesd on Gabor transform is presented according to the characteristics of time-frequency analysis in Gabor transform. The similarity (i.e. distance) for echo signal and the Gabor transform window function is modeled. Time Of Flight (TOF) and Center Frequency (CF) of echo signal are estimated by translating model for solving the minimum into solving the maximum of Gabor transform coefficient modulus. Finally, the CRLB is derived to evaluate the performance of the algorithm. The Monte Carlo simulation and experimental results show that the proposed method is efficient and successful. The estimation of single echo or overlapping echoes obtains high accuracy even in low signal to noise ratio. The Mean Square Error (MSE) of estimation achieves CRLB at high SNR, and is close to CRLB, even in low signal to noise ratio.
In non-destructive testing, ultrasonic echo is often an overlapping multi-echoes signal with noise. A time- frequency estimation algorithm for ultrasonic echo signal baesd on Gabor transform is presented according to the characteristics of time-frequency analysis in Gabor transform. The similarity (i.e. distance) for echo signal and the Gabor transform window function is modeled. Time Of Flight (TOF) and Center Frequency (CF) of echo signal are estimated by translating model for solving the minimum into solving the maximum of Gabor transform coefficient modulus. Finally, the CRLB is derived to evaluate the performance of the algorithm. The Monte Carlo simulation and experimental results show that the proposed method is efficient and successful. The estimation of single echo or overlapping echoes obtains high accuracy even in low signal to noise ratio. The Mean Square Error (MSE) of estimation achieves CRLB at high SNR, and is close to CRLB, even in low signal to noise ratio.
2013, 35(3): 658-664.
doi: 10.3724/SP.J.1146.2012.00896
Abstract:
Nonsinusoidal orthogonal modulation signal in time-domain has high peak-to-average power ratio, so it is vulnerable to the nonlinearity of power amplifier. To relieve this affection, a non-iterative look-up table predistortion method based on waveform training is presented which can obtain the amplitude and phase modification value fast and corretly. For the reason that the predistortion accuracy is directly associated with the chosen quantization method, a compressing quantization method is presented. The characteristic of power amplifier and the distribution characteristic of signal amplitude are taken considered in this method. The distortion characteristic in band and out of band of signal before and after imploying the presented predistortion method is simulated and compared with the method with equal space. The results show that the presented method can effectively improve the power spectrum of the signal and the bit error rate performance of system, and minish the intermodulated power of signal.
Nonsinusoidal orthogonal modulation signal in time-domain has high peak-to-average power ratio, so it is vulnerable to the nonlinearity of power amplifier. To relieve this affection, a non-iterative look-up table predistortion method based on waveform training is presented which can obtain the amplitude and phase modification value fast and corretly. For the reason that the predistortion accuracy is directly associated with the chosen quantization method, a compressing quantization method is presented. The characteristic of power amplifier and the distribution characteristic of signal amplitude are taken considered in this method. The distortion characteristic in band and out of band of signal before and after imploying the presented predistortion method is simulated and compared with the method with equal space. The results show that the presented method can effectively improve the power spectrum of the signal and the bit error rate performance of system, and minish the intermodulated power of signal.
2013, 35(3): 665-670.
doi: 10.3724/SP.J.1146.2012.00860
Abstract:
Channel estimation which based on Compressed Sensing (CS) can achieve the purpose of reducing pilots, but in the transformation of channel matrix from frequency-time domain to delay-Doppler sparse domain exists spectral leakage phenomenon which affects the sparsity of the channel and the Mean Squared Error (MSE) performance of estimation. For this, this paper studies the sparsity of the channel and a compressed channel estimation algorithm which optimized the sparsity by time domain windowing is proposed. With time domain windowing, the proposed algorithm restrains the leakage of Doppler domain which is caused by discretization and truncation, then the measurement matrix is designed. By this method, the sparsity of the delay-Doppler domain channel is enhanced and the more accurate sparse channel matrix is reconstructed. The channel estimation performance is improved. Simulation results show that with the signal-to-noise ratio increasing, windowed CS algorithm improves effectively the performance of channel estimation compared with no windows CS algorithm.
Channel estimation which based on Compressed Sensing (CS) can achieve the purpose of reducing pilots, but in the transformation of channel matrix from frequency-time domain to delay-Doppler sparse domain exists spectral leakage phenomenon which affects the sparsity of the channel and the Mean Squared Error (MSE) performance of estimation. For this, this paper studies the sparsity of the channel and a compressed channel estimation algorithm which optimized the sparsity by time domain windowing is proposed. With time domain windowing, the proposed algorithm restrains the leakage of Doppler domain which is caused by discretization and truncation, then the measurement matrix is designed. By this method, the sparsity of the delay-Doppler domain channel is enhanced and the more accurate sparse channel matrix is reconstructed. The channel estimation performance is improved. Simulation results show that with the signal-to-noise ratio increasing, windowed CS algorithm improves effectively the performance of channel estimation compared with no windows CS algorithm.
2013, 35(3): 671-676.
doi: 10.3724/SP.J.1146.2012.00995
Abstract:
The digital channelizer processing on-board is adopted by the broadband flexible transponder, which is the next generation communication satellite payload. This payload can flexibly transpond signals within any frequency band and any bandwidth, which can not be well solving by the traditional payloads. Based on analyzing performance of the modulated filter banks in flexible transponder, an algorithm for designing prototype filter with Near Perfect Reconstruction (NPR) is proposed through converting the filter banks to the design of prototype filter. This method include two step: Firstly, the low order two-subchannel filter h(2)(n) and the real symmetric FIR filter g(n) are designed using the Parks-McClellan algorithm. Then the desired NPR prototype filter h(M)(n) is calculated through upsampling non-zero values to the coefficients of h(2)(n) with a factor of M and factor-2 downsampling according filter g(n). So the proposed algorithm can efficiently avoid the performance loss caused by image components with Interpolated Finite Impulse Response (IFIR) approach. The simulation results prove that the modulated filter banks designed by the proposed algorithm present better performance than conventional methods.
The digital channelizer processing on-board is adopted by the broadband flexible transponder, which is the next generation communication satellite payload. This payload can flexibly transpond signals within any frequency band and any bandwidth, which can not be well solving by the traditional payloads. Based on analyzing performance of the modulated filter banks in flexible transponder, an algorithm for designing prototype filter with Near Perfect Reconstruction (NPR) is proposed through converting the filter banks to the design of prototype filter. This method include two step: Firstly, the low order two-subchannel filter h(2)(n) and the real symmetric FIR filter g(n) are designed using the Parks-McClellan algorithm. Then the desired NPR prototype filter h(M)(n) is calculated through upsampling non-zero values to the coefficients of h(2)(n) with a factor of M and factor-2 downsampling according filter g(n). So the proposed algorithm can efficiently avoid the performance loss caused by image components with Interpolated Finite Impulse Response (IFIR) approach. The simulation results prove that the modulated filter banks designed by the proposed algorithm present better performance than conventional methods.
2013, 35(3): 677-682.
doi: 10.3724/SP.J.1146.2012.00820
Abstract:
Turbo equalization is the effective method to overcome multipath fading and eliminate Inter-Symbol Interference (ISI) for underwater acoustic coherent communications. In practice, turbo equalization estimates information of time-varying and multi-path channel. In order to improve the performance of channel and phase estimation, an algorithm of combining soft iterative channel and phase estimation for turbo equalization is proposed based on channel mode of time-varying transversal filter and phase rotation. The algorithm adopts soft iterative fast self-optimized LMS to get transversal filter coefficients vector for every symbol and combing it and second-order phase-locked loop for optimization. Through simulation, the proposed algorithm is better than conventional hard decision channel and phase estimation algorithm. As to performance of phase estimation, the proposed algorithm is better than soft iterative channel estimation raised in other references. During the sea experiment, underwater acoustic communication distance is 5 m, approximately in vertical direction and the fuctuation period of receiver array is 10 s. The algorithm is adopted for single-channel Turbo equalization on sea experiment data, which can output error-free symbols. The result verifies the advantage of all of the algorithms in the soft channel estimation.
Turbo equalization is the effective method to overcome multipath fading and eliminate Inter-Symbol Interference (ISI) for underwater acoustic coherent communications. In practice, turbo equalization estimates information of time-varying and multi-path channel. In order to improve the performance of channel and phase estimation, an algorithm of combining soft iterative channel and phase estimation for turbo equalization is proposed based on channel mode of time-varying transversal filter and phase rotation. The algorithm adopts soft iterative fast self-optimized LMS to get transversal filter coefficients vector for every symbol and combing it and second-order phase-locked loop for optimization. Through simulation, the proposed algorithm is better than conventional hard decision channel and phase estimation algorithm. As to performance of phase estimation, the proposed algorithm is better than soft iterative channel estimation raised in other references. During the sea experiment, underwater acoustic communication distance is 5 m, approximately in vertical direction and the fuctuation period of receiver array is 10 s. The algorithm is adopted for single-channel Turbo equalization on sea experiment data, which can output error-free symbols. The result verifies the advantage of all of the algorithms in the soft channel estimation.
2013, 35(3): 683-688.
doi: 10.3724/SP.J.1146.2012.00948
Abstract:
Frequency-domain equalization for single-carrier transmission systems provides an attractive design alterative to time domain equalization because of more efficient calculation. In this paper an Iterative Block Decision Feedback Equalization (IB-DFE) scheme for communication systems over UnderWater Acoustic (UWA) channels with long delay and Doppler frequency shifts is researched and a frequency domain channel estimation algorithm Jointing iterative Equalization and Channel Estimation (JECE) is proposed. The performance of JECE algorithm is evaluated and compared. The influences of different data parameters on the performance of IB-DFE are analyzed and the computational complexities of IB-DFE are calculated. Numerical results show that the proposed JECE algorithm is more adaptive to the time-varying UWA channels. Compared with Time Domain Decision Feedback Equalization (TD-DFE), IB-DFE has more than 3.6 dB signal-to-noise ratio gain and 19% complexity gain and these gains will extend as the increasing of delay and Doppler frequency shifts of UWA channels.
Frequency-domain equalization for single-carrier transmission systems provides an attractive design alterative to time domain equalization because of more efficient calculation. In this paper an Iterative Block Decision Feedback Equalization (IB-DFE) scheme for communication systems over UnderWater Acoustic (UWA) channels with long delay and Doppler frequency shifts is researched and a frequency domain channel estimation algorithm Jointing iterative Equalization and Channel Estimation (JECE) is proposed. The performance of JECE algorithm is evaluated and compared. The influences of different data parameters on the performance of IB-DFE are analyzed and the computational complexities of IB-DFE are calculated. Numerical results show that the proposed JECE algorithm is more adaptive to the time-varying UWA channels. Compared with Time Domain Decision Feedback Equalization (TD-DFE), IB-DFE has more than 3.6 dB signal-to-noise ratio gain and 19% complexity gain and these gains will extend as the increasing of delay and Doppler frequency shifts of UWA channels.
2013, 35(3): 689-695.
doi: 10.3724/SP.J.1146.2012.00811
Abstract:
Search patterns have an important influence on both searching speed and encoding quality of Block- based Motion estimation Algorithms (BMA). A fast block-based motion estimation algorithm utilizing a multi- pattern switching method is proposed, during which process a halfway stop technology and a selective search method are used to promote encoding speed. The algorithm is named Diamond-Hexagon-Square (DHS) algorithm, for a small diamond pattern is used as the initial search pattern, followed by a hexagon pattern, and a square search pattern is utilized at last for refinement. Experimental results indicate that the DHS algorithm is suitable for various video sequences with different motion contents (slow, medium and fast). It is faster than the small DIAmond search algorithm (DIA), HEXagon search algorithm (HEX), a Exhausted Search Algorithm (ESA) and Unsymmetrical-cross Multi-Hexagon-grid Search (UMHexagonS) algorithm with no obvious rate-distortion performance depravation.
Search patterns have an important influence on both searching speed and encoding quality of Block- based Motion estimation Algorithms (BMA). A fast block-based motion estimation algorithm utilizing a multi- pattern switching method is proposed, during which process a halfway stop technology and a selective search method are used to promote encoding speed. The algorithm is named Diamond-Hexagon-Square (DHS) algorithm, for a small diamond pattern is used as the initial search pattern, followed by a hexagon pattern, and a square search pattern is utilized at last for refinement. Experimental results indicate that the DHS algorithm is suitable for various video sequences with different motion contents (slow, medium and fast). It is faster than the small DIAmond search algorithm (DIA), HEXagon search algorithm (HEX), a Exhausted Search Algorithm (ESA) and Unsymmetrical-cross Multi-Hexagon-grid Search (UMHexagonS) algorithm with no obvious rate-distortion performance depravation.
2013, 35(3): 696-702.
doi: 10.3724/SP.J.1146.2012.00901
Abstract:
In modern wireless networks, Adaptive Modulation and Coding (AMC) is one of effective technologies adopted in physical layer to overcome complex wireless channel and improve quality of service of system. In this paper, the relation between wireless channel service process and traffic flow and AMC adopted in physical layer are analyzed, the wireless channel service process is modeled; and an equivalent burstiness characteristic AMC wireless channel modeling method is proposed. The burstiness character of AMC wireless channel and the feasibility and effectiveness of the proposed modeling method are proved with numerical analyses.
In modern wireless networks, Adaptive Modulation and Coding (AMC) is one of effective technologies adopted in physical layer to overcome complex wireless channel and improve quality of service of system. In this paper, the relation between wireless channel service process and traffic flow and AMC adopted in physical layer are analyzed, the wireless channel service process is modeled; and an equivalent burstiness characteristic AMC wireless channel modeling method is proposed. The burstiness character of AMC wireless channel and the feasibility and effectiveness of the proposed modeling method are proved with numerical analyses.
2013, 35(3): 703-708.
doi: 10.3724/SP.J.1146.2012.00973
Abstract:
Since the traditional network architectures faced the problems of structural rigidity, functional simplification and poor controllability, the reconfigurable network is designed as a new kind of network architecture. In the reconfigurable network, traffic clustering is the key mechanism. By detailed analysis of the requirement of traffic clustering and the characteristics of clustering algorithms, Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm is chosen for traffic clustering. However, this algorithm has a poor performance in identifying clusters of arbitrary shapes. In order to solve this issue, an advanced algorithm named Advanced Split BIRCH (AS-BIRCH) is designed. Experimental results demonstrate the effectiveness of AS-BIRCH in traffic clustering.
Since the traditional network architectures faced the problems of structural rigidity, functional simplification and poor controllability, the reconfigurable network is designed as a new kind of network architecture. In the reconfigurable network, traffic clustering is the key mechanism. By detailed analysis of the requirement of traffic clustering and the characteristics of clustering algorithms, Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm is chosen for traffic clustering. However, this algorithm has a poor performance in identifying clusters of arbitrary shapes. In order to solve this issue, an advanced algorithm named Advanced Split BIRCH (AS-BIRCH) is designed. Experimental results demonstrate the effectiveness of AS-BIRCH in traffic clustering.
2013, 35(3): 709-714.
doi: 10.3724/SP.J.1146.2012.00890
Abstract:
To increase the end-to-end system efficiency with a flexible QoS (Quality of Service) class mapping scheme for transporting multimedia traffic over heterogeneous networks, by analyzing current QoS class mapping schemes which lack flexibility without considering users QoE (Quality of Experience), an analytical model is developed for end-to-end QoS provisioning based on network calculus theory. Based on this model, the paper derives an Elastic QoS Class Mapping Method (EQCMM) which considers users QoE. With EQCMM, the result of QoS class mapping can be adjusted elastically according to available network resources, in order to achieve higher end-to-end system efficiency. Finally, simulation results are given to demonstrate the effectiveness of the proposed method.
To increase the end-to-end system efficiency with a flexible QoS (Quality of Service) class mapping scheme for transporting multimedia traffic over heterogeneous networks, by analyzing current QoS class mapping schemes which lack flexibility without considering users QoE (Quality of Experience), an analytical model is developed for end-to-end QoS provisioning based on network calculus theory. Based on this model, the paper derives an Elastic QoS Class Mapping Method (EQCMM) which considers users QoE. With EQCMM, the result of QoS class mapping can be adjusted elastically according to available network resources, in order to achieve higher end-to-end system efficiency. Finally, simulation results are given to demonstrate the effectiveness of the proposed method.
2013, 35(3): 715-720.
doi: 10.3724/SP.J.1146.2012.00935
Abstract:
To deal with dynamic routing in wireless sensor networks and keep the residual-energy balance of each sensor node, an opportunistic routing algorithm is proposed. At first, based on the second law of thermodynamics and the model of entropy, an opportunistic entropy is proposed to describe each sensor nodes real-time status in wireless sensor network, which the nodes energy and communication distance to the sink node are mainly considered. Then, based on the opportunistic entropy and the Ant Colony Optimization (ACO), an opportunistic routing protocol for wireless sensor network, called ACO for Time Dependent Opportunistic-routing Protocol (ATDOP), is proposed in this paper. In ATDOP, the node with the lowest opportunistic entropy in the sending nodes neighbors, is chosen to store and relay packets at each hop. Finally, the simulation results show that ATDOP performs better on successful delivery ratio, throughput and energy overhead as compared with traditional routing protocols.
To deal with dynamic routing in wireless sensor networks and keep the residual-energy balance of each sensor node, an opportunistic routing algorithm is proposed. At first, based on the second law of thermodynamics and the model of entropy, an opportunistic entropy is proposed to describe each sensor nodes real-time status in wireless sensor network, which the nodes energy and communication distance to the sink node are mainly considered. Then, based on the opportunistic entropy and the Ant Colony Optimization (ACO), an opportunistic routing protocol for wireless sensor network, called ACO for Time Dependent Opportunistic-routing Protocol (ATDOP), is proposed in this paper. In ATDOP, the node with the lowest opportunistic entropy in the sending nodes neighbors, is chosen to store and relay packets at each hop. Finally, the simulation results show that ATDOP performs better on successful delivery ratio, throughput and energy overhead as compared with traditional routing protocols.
2013, 35(3): 721-727.
doi: 10.3724/SP.J.1146.2012.01013
Abstract:
Considering the disadvantage of the high complexity and ignoring signals structural sparsity in A* Orthogonal Matching Pursuit (A*OMP) algorithm, a block A*OMP algorithm is proposed for block-sparse signals, and it is improved to solve the joint reconstruction problem for multiple signals in distributed compressed sensing. In the proposed algorithm, the single atom is replaced by a block that is composed of several atoms, and the sparsity is replaced by the maximum length of all the paths on the search tree when calculating the path cost. Then, on the basis of block A*OMP algorithm, a block A*OMP algorithm for Multiple Measurement Vector (MMV) problem is presented by projecting all blocks onto the residual matrix and selecting the block with the smallest projection error as a new node. With this algorithm, the temperature signals which are measured by sensors in the adjacent region are jointly reconstructed perfectly. Experiments show that the reconstruction performance of this algorithm outperform Orthogonal Matching Pursuit for MMV (OMPMMV) algorithm.
Considering the disadvantage of the high complexity and ignoring signals structural sparsity in A* Orthogonal Matching Pursuit (A*OMP) algorithm, a block A*OMP algorithm is proposed for block-sparse signals, and it is improved to solve the joint reconstruction problem for multiple signals in distributed compressed sensing. In the proposed algorithm, the single atom is replaced by a block that is composed of several atoms, and the sparsity is replaced by the maximum length of all the paths on the search tree when calculating the path cost. Then, on the basis of block A*OMP algorithm, a block A*OMP algorithm for Multiple Measurement Vector (MMV) problem is presented by projecting all blocks onto the residual matrix and selecting the block with the smallest projection error as a new node. With this algorithm, the temperature signals which are measured by sensors in the adjacent region are jointly reconstructed perfectly. Experiments show that the reconstruction performance of this algorithm outperform Orthogonal Matching Pursuit for MMV (OMPMMV) algorithm.
2013, 35(3): 728-734.
doi: 10.3724/SP.J.1146.2012.00970
Abstract:
Due to marine dynamics, sensor nodes of Underwater Wireless Sensor Networks (UWSN) are always in a state of constant motion. Together with low propagation speed of acoustic signal, nodes mobility brings challenge to UWSN MAC. A motion model of underwater node is established to solve the multiple access problems for underwater mobile networks. Based on AR mobility prediction algorithm, the negative effect of space-time uncertainty can be reduced and the probability that sending information arrives in the reservation slot can be improved. The simulation results show that AR(5) algorithm can decrease 74.8% of delay detection error. A new contention-based MAC protocol: P-MAC is proposed. NS-2 simulation results show that P-MAC can get a 10% -15% higher Packet Receptipn Rate (PRR) under the circumstance of wave motion.
Due to marine dynamics, sensor nodes of Underwater Wireless Sensor Networks (UWSN) are always in a state of constant motion. Together with low propagation speed of acoustic signal, nodes mobility brings challenge to UWSN MAC. A motion model of underwater node is established to solve the multiple access problems for underwater mobile networks. Based on AR mobility prediction algorithm, the negative effect of space-time uncertainty can be reduced and the probability that sending information arrives in the reservation slot can be improved. The simulation results show that AR(5) algorithm can decrease 74.8% of delay detection error. A new contention-based MAC protocol: P-MAC is proposed. NS-2 simulation results show that P-MAC can get a 10% -15% higher Packet Receptipn Rate (PRR) under the circumstance of wave motion.
2013, 35(3): 735-741.
doi: 10.3724/SP.J.1146.2012.00930
Abstract:
For meeting the high security and high efficiency of cryptography schemes in Cloud Computing, a security enhanced cryptographic mode of operation named Cipher FeedBack one Block one Key (CFB-BK) is proposed, and this mode is implemented based on the combination of mathematical cryptography and optical cryptography. Optical cryptography module encrypts (or decrypts) data blocks in a one block one key way. Moreover, it provides its ciphertext as a feedback to the mathematical cryptography module, which uses it to generate keys for the next block encryption (or decryption) of the optical cryptography module. Security analysis shows that the only possible attack in the area of cryptography for the proposed scheme is exhaustive attack, indicating that no adversary could get a significant advantage against the scheme without spending a huge amount of recourses and time. Efficiency analysis shows that the scheme implementing CFB-BK mode works much faster than the existing modes implemented based on mathematical cryptography.
For meeting the high security and high efficiency of cryptography schemes in Cloud Computing, a security enhanced cryptographic mode of operation named Cipher FeedBack one Block one Key (CFB-BK) is proposed, and this mode is implemented based on the combination of mathematical cryptography and optical cryptography. Optical cryptography module encrypts (or decrypts) data blocks in a one block one key way. Moreover, it provides its ciphertext as a feedback to the mathematical cryptography module, which uses it to generate keys for the next block encryption (or decryption) of the optical cryptography module. Security analysis shows that the only possible attack in the area of cryptography for the proposed scheme is exhaustive attack, indicating that no adversary could get a significant advantage against the scheme without spending a huge amount of recourses and time. Efficiency analysis shows that the scheme implementing CFB-BK mode works much faster than the existing modes implemented based on mathematical cryptography.
2013, 35(3): 742-748.
doi: 10.3724/SP.J.1146.2012.00878
Abstract:
Model construction is the basis of model checking. State explosion can not be avoided during building model for microcontroller code. Because the state number of generated model is related to code size, the number of state can be reduced through simplifying microcontroller code. An algorithm of sensitive position identification with subroutine summary information is proposed, based on concepts of sensitive variable and sensitive position. Sensitive variables are extracted from verified properties and used to identify sensitive positions. Then model is constructed from code corresponding to sensitive positions. Experimental results show that the problem of state explosion can be effectively alleviated through the proposed method.
Model construction is the basis of model checking. State explosion can not be avoided during building model for microcontroller code. Because the state number of generated model is related to code size, the number of state can be reduced through simplifying microcontroller code. An algorithm of sensitive position identification with subroutine summary information is proposed, based on concepts of sensitive variable and sensitive position. Sensitive variables are extracted from verified properties and used to identify sensitive positions. Then model is constructed from code corresponding to sensitive positions. Experimental results show that the problem of state explosion can be effectively alleviated through the proposed method.
2013, 35(3): 749-753.
doi: 10.3724/SP.J.1146.2012.00915
Abstract:
In order to figure out the distortion of radiation pattern, which is caused by backward leakage, second contamination and output boundary truncation, a novel Local Waveguide Port (LWP) technique based on FDTD technique is proposed in this paper. And the detail of boundary conditions and boundary parameters are given for the method. At the end of this paper, this LWP technique is utilized to analyze discontinuous rectangular waveguide, micro-strip antenna and slot antenna array excited by local waveguide port. The calculation results show that the high accuracy can be obtained in the discontinuous problem and the much better result can be achieved in the radiant problem. Consequently, this method can be used in both discontinuous problem and radiant problem.
In order to figure out the distortion of radiation pattern, which is caused by backward leakage, second contamination and output boundary truncation, a novel Local Waveguide Port (LWP) technique based on FDTD technique is proposed in this paper. And the detail of boundary conditions and boundary parameters are given for the method. At the end of this paper, this LWP technique is utilized to analyze discontinuous rectangular waveguide, micro-strip antenna and slot antenna array excited by local waveguide port. The calculation results show that the high accuracy can be obtained in the discontinuous problem and the much better result can be achieved in the radiant problem. Consequently, this method can be used in both discontinuous problem and radiant problem.
2013, 35(3): 754-758.
doi: 10.3724/SP.J.1146.2012.00923
Abstract:
In recent years, the global wind power installed capacity is growing exponentially. However, current Moving Target Detector (MTD) of primary surveillance radar can not suppress the wind farm clutter. The non-zero frequency components of the wind farm clutter may produce target losing and higher false alarm. A novel method for radar target detection in wind farm clutter is proposed. The method sets a wind farm clutter suppressor before current MTD. First, the center Doppler frequency of each range cell is estimated and the spectral center of each range cell is shifted to the zero frequency. Then, target echo is removed with the method similar to canceling stationary clutter. The wind farm clutter with rather wide spectral width still reserves most of energy after canceller process. Finally, the clutter units detected by Constant False Alarm Rate (CFAR) are eliminated. The proposed method prevented the obvious deterioration of current MTD detection due to clutter energy spreading into non-zero Doppler filters. The experimental results indicate that the method conqueres the problem of masked target and decreased false alarms effectively are not sensitive to the power of the wind farm clutter.
In recent years, the global wind power installed capacity is growing exponentially. However, current Moving Target Detector (MTD) of primary surveillance radar can not suppress the wind farm clutter. The non-zero frequency components of the wind farm clutter may produce target losing and higher false alarm. A novel method for radar target detection in wind farm clutter is proposed. The method sets a wind farm clutter suppressor before current MTD. First, the center Doppler frequency of each range cell is estimated and the spectral center of each range cell is shifted to the zero frequency. Then, target echo is removed with the method similar to canceling stationary clutter. The wind farm clutter with rather wide spectral width still reserves most of energy after canceller process. Finally, the clutter units detected by Constant False Alarm Rate (CFAR) are eliminated. The proposed method prevented the obvious deterioration of current MTD detection due to clutter energy spreading into non-zero Doppler filters. The experimental results indicate that the method conqueres the problem of masked target and decreased false alarms effectively are not sensitive to the power of the wind farm clutter.
2013, 35(3): 759-762.
doi: 10.3724/SP.J.1146.2012.01177
Abstract:
Based on Wyner-Ziv (WZ) codec structure, a four-channel distributed codec methods for Bayer pattern image is proposed. First, Bayer pattern image is separated and converted into four component image, and each component image is transformed into frequency domain by Discreted Cosine Transform (DCT). According to the convergence of Lagrangian cost function, an universal optimal quantizier is designed by Lloyd iterative algorithm. The quantized DCT coefficients in each channel are encoded independently by employing Slepian-Wolf (SW) coder. At the side of decoder, the side information is generated from illumination component image and available for the SW decoder and the inverse quantizier. Finally, the Bayer pattern image is rebuilt at the side of decoder. The experiments show that the proposed method can improve rate-distortion performance of Bayer pattern image codec at high rate.
Based on Wyner-Ziv (WZ) codec structure, a four-channel distributed codec methods for Bayer pattern image is proposed. First, Bayer pattern image is separated and converted into four component image, and each component image is transformed into frequency domain by Discreted Cosine Transform (DCT). According to the convergence of Lagrangian cost function, an universal optimal quantizier is designed by Lloyd iterative algorithm. The quantized DCT coefficients in each channel are encoded independently by employing Slepian-Wolf (SW) coder. At the side of decoder, the side information is generated from illumination component image and available for the SW decoder and the inverse quantizier. Finally, the Bayer pattern image is rebuilt at the side of decoder. The experiments show that the proposed method can improve rate-distortion performance of Bayer pattern image codec at high rate.
