ISSN 
2019 Vol. 48, No. 3
2019, 48(3): 321-325.
doi: 10.3969/j.issn.1001-0548.2019.03.001
Abstract:
This paper deals with the tracking problem of multiple human targets hidden behind walls using through-wall-imaging (TWI) radar. We propose a mean-shift based multi-target tracking algorithm in dzkjdxxb-48-3-321 domain. Firstly, an adaptive method of target template construction based on biaxial projection is presented. Secondly, the mean-shift algorithm is used to update the trajectories, and the different amplitudes is weighted in order to make the iterative process converge to the center of the target. The M/N logic is finally considered to manage trajectories. The proposed algorithm is evaluated by the experiments.
This paper deals with the tracking problem of multiple human targets hidden behind walls using through-wall-imaging (TWI) radar. We propose a mean-shift based multi-target tracking algorithm in dzkjdxxb-48-3-321 domain. Firstly, an adaptive method of target template construction based on biaxial projection is presented. Secondly, the mean-shift algorithm is used to update the trajectories, and the different amplitudes is weighted in order to make the iterative process converge to the center of the target. The M/N logic is finally considered to manage trajectories. The proposed algorithm is evaluated by the experiments.
2019, 48(3): 326-330.
doi: 10.3969/j.issn.1001-0548.2019.03.002
Abstract:
A detection method of radar small target in a strong clutter environment is proposed. On the basis of singular value decomposition theory, the first order and second order difference spectrum of singular values is used to select the singular values in various combination. Radar echo signal is decomposed into different compositions by inverse singular value transformation, thus realizing the clutter suppression and small target highlights. The experiment shows that the method can effectively suppress the strong clutter and improve the signal to noise ratio 7dB of small targets.
A detection method of radar small target in a strong clutter environment is proposed. On the basis of singular value decomposition theory, the first order and second order difference spectrum of singular values is used to select the singular values in various combination. Radar echo signal is decomposed into different compositions by inverse singular value transformation, thus realizing the clutter suppression and small target highlights. The experiment shows that the method can effectively suppress the strong clutter and improve the signal to noise ratio 7dB of small targets.
2019, 48(3): 331-335.
doi: 10.3969/j.issn.1001-0548.2019.03.003
Abstract:
In ultra-wideband (UWB)-based indoor positioning, the number of base stations used for location calculating is always affected by non-line-of-sight (NLOS) propagation or sheltering, and thus leading to no solution of location equations. This paper proposes a precise indoor localization algorithm based on received signal strength indicator (RSSI) and UWB ranging techniques. The multiple RSSI measurements of wireless local area network (WLAN) is transformed into the corresponding distances, which can be used to improve the positioning accuracy of the UWB. Our technique can localize targets by minimizing the positioning errors with only one UWB base station. Compared with the existing least squares (LS) and maximization likelihood (ML) algorithms, our method achieves better performances in arbitrary structure of base stations with low computational complexity.
In ultra-wideband (UWB)-based indoor positioning, the number of base stations used for location calculating is always affected by non-line-of-sight (NLOS) propagation or sheltering, and thus leading to no solution of location equations. This paper proposes a precise indoor localization algorithm based on received signal strength indicator (RSSI) and UWB ranging techniques. The multiple RSSI measurements of wireless local area network (WLAN) is transformed into the corresponding distances, which can be used to improve the positioning accuracy of the UWB. Our technique can localize targets by minimizing the positioning errors with only one UWB base station. Compared with the existing least squares (LS) and maximization likelihood (ML) algorithms, our method achieves better performances in arbitrary structure of base stations with low computational complexity.
2019, 48(3): 336-339.
doi: 10.3969/j.issn.1001-0548.2019.03.004
Abstract:
Wireless broadcast environment always affects the transmission quality in wireless sensor networks. In order to improve the SNR and communication rate, an opportunistic scheduling algorithm is adopted to schedule the sensor nodes which are in the good channel states. Multiple antennas are equipped at central node, the other nodes are equipped with single antenna to save energy and volume, which is a multiple input and single output multi-user system. The communication is divided into two stages:at first, the random unitary beamforming is generated by multi-antennas at the pilot stage; secondly, after receiving the pilot information, the sensor node determines whether to enter the communication state according to the preset threshold and then selects the corresponding random beam. The properly truncated threshold is given by Monte Carlo method through communication rate comparison under different situations. The results shows that the SNR and communication rate are both improved by the opportunistic scheduling.
Wireless broadcast environment always affects the transmission quality in wireless sensor networks. In order to improve the SNR and communication rate, an opportunistic scheduling algorithm is adopted to schedule the sensor nodes which are in the good channel states. Multiple antennas are equipped at central node, the other nodes are equipped with single antenna to save energy and volume, which is a multiple input and single output multi-user system. The communication is divided into two stages:at first, the random unitary beamforming is generated by multi-antennas at the pilot stage; secondly, after receiving the pilot information, the sensor node determines whether to enter the communication state according to the preset threshold and then selects the corresponding random beam. The properly truncated threshold is given by Monte Carlo method through communication rate comparison under different situations. The results shows that the SNR and communication rate are both improved by the opportunistic scheduling.
2019, 48(3): 340-344.
doi: 10.3969/j.issn.1001-0548.2019.03.005
Abstract:
This paper presents an improved control strategy for tracking system based on photoelectric sensors. Taking the sun-tracking system as an example, the new three-quadrant position sensor is used to replace the original four-quadrant sensor for finding directions, which reduces the system cost and difficulties of controlling. The improved system control strategy uses the differential effect to reduce the analog quantities to two, corresponding respectively to the two-dimension control of the horizontal and vertical directions. Compared with the original control strategy, the proposed control logic is simpler, thus the real time and reliability can also be guaranteed. Tests explain that the system can realize the tracking of the point light source in the whole airspace and replace the existing four-quadrant sun-tracking system. By replacing the terminal sensor, the improved control strategy can also be used to track electromagnetic wave sources in space, which has great application value in both military and civil fields.
This paper presents an improved control strategy for tracking system based on photoelectric sensors. Taking the sun-tracking system as an example, the new three-quadrant position sensor is used to replace the original four-quadrant sensor for finding directions, which reduces the system cost and difficulties of controlling. The improved system control strategy uses the differential effect to reduce the analog quantities to two, corresponding respectively to the two-dimension control of the horizontal and vertical directions. Compared with the original control strategy, the proposed control logic is simpler, thus the real time and reliability can also be guaranteed. Tests explain that the system can realize the tracking of the point light source in the whole airspace and replace the existing four-quadrant sun-tracking system. By replacing the terminal sensor, the improved control strategy can also be used to track electromagnetic wave sources in space, which has great application value in both military and civil fields.
2019, 48(3): 345-350, 373.
doi: 10.3969/j.issn.1001-0548.2019.03.006
Abstract:
For the construction of importance density function (IDF) of Gaussian particle filter, recursive update Gaussian filter (RUGF) which can effectively overcome the limitation of linear minimum mean square error criterion, updates the target state incrementally based on the gradient of nonlinear measurement function. Consequently, the posterior state estimation that is closer to the real distribution is obtained, but non-positive definite state covariance matrix will lead to recursive interruption. To solve this problem, the square-root implementation strategy of RUGF is firstly analyzed and then square-root recursive update Gaussian filter (SR-RUGF) is implemented by using cubature Kalman filter. Based on that, SR-RUGF is used to construct IDF for Gaussian particle filter. Simulation results demonstrate that the proposed algorithm can effectively solve the recursive interruption problem and obtain estimation result with higher accuracy.
For the construction of importance density function (IDF) of Gaussian particle filter, recursive update Gaussian filter (RUGF) which can effectively overcome the limitation of linear minimum mean square error criterion, updates the target state incrementally based on the gradient of nonlinear measurement function. Consequently, the posterior state estimation that is closer to the real distribution is obtained, but non-positive definite state covariance matrix will lead to recursive interruption. To solve this problem, the square-root implementation strategy of RUGF is firstly analyzed and then square-root recursive update Gaussian filter (SR-RUGF) is implemented by using cubature Kalman filter. Based on that, SR-RUGF is used to construct IDF for Gaussian particle filter. Simulation results demonstrate that the proposed algorithm can effectively solve the recursive interruption problem and obtain estimation result with higher accuracy.
2019, 48(3): 351-355.
doi: 10.3969/j.issn.1001-0548.2019.03.007
Abstract:
Heterogeneous network with full-dimensional (FD) multiple-input multiple-output (MIMO) can effectively improve the transmission rate by developing vertical dimensions; however, it can cause the signal leakage and decrease the system's physical layer security performances. In this paper, using the beamforming to eliminate the interference in the downlink of the heterogeneous network, we propose an analysis method with stochastic geometric theory to obtain the closed-form expression of the secure connection probability on the legitimate users. The secure connection probability is related to the density of heterogeneous network small cell base stations (BSs). Simulation results verify the correctness of the safety connection probability, and determine the density value of heterogeneous network BS in a given scenario.
Heterogeneous network with full-dimensional (FD) multiple-input multiple-output (MIMO) can effectively improve the transmission rate by developing vertical dimensions; however, it can cause the signal leakage and decrease the system's physical layer security performances. In this paper, using the beamforming to eliminate the interference in the downlink of the heterogeneous network, we propose an analysis method with stochastic geometric theory to obtain the closed-form expression of the secure connection probability on the legitimate users. The secure connection probability is related to the density of heterogeneous network small cell base stations (BSs). Simulation results verify the correctness of the safety connection probability, and determine the density value of heterogeneous network BS in a given scenario.
2019, 48(3): 356-360.
doi: 10.3969/j.issn.1001-0548.2019.03.008
Abstract:
Recently, atom-based microwave (MW) measurement has inspired great interest because of its potential ability to link the MW quantity with the international system of units (SI) second. The frequency has the highest measurement accuracy among all physical quantities, implying a great potential of atomic MW measurement. At present, the main factor limiting the measurement accuracy arises from atomic vapor-cell itself. In order to evaluate the effects of the vapor-cell on atom-based MW measurements, the structure parameters and permittivity of vapor-cell are firstly estimated in this paper. As a demonstration, the complex permittivity of a cylindrical vapor-cell is measured and evaluated through MW cavity perturbation technique at S and X bands. Finally, various methods used for the measurement of permittivity of vapor-cell are briefly discussed.
Recently, atom-based microwave (MW) measurement has inspired great interest because of its potential ability to link the MW quantity with the international system of units (SI) second. The frequency has the highest measurement accuracy among all physical quantities, implying a great potential of atomic MW measurement. At present, the main factor limiting the measurement accuracy arises from atomic vapor-cell itself. In order to evaluate the effects of the vapor-cell on atom-based MW measurements, the structure parameters and permittivity of vapor-cell are firstly estimated in this paper. As a demonstration, the complex permittivity of a cylindrical vapor-cell is measured and evaluated through MW cavity perturbation technique at S and X bands. Finally, various methods used for the measurement of permittivity of vapor-cell are briefly discussed.
2019, 48(3): 361-365.
doi: 10.3969/j.issn.1001-0548.2019.03.009
Abstract:
With the widely deployment of information techniques in smart grid, it is quite important to automatically detect the bad data, e.g., malicious injection data and unfunctional sensor data, from daily observations. In this paper, we propose a novel approach for bad data detection in smart grid based on multi-view low-rank analysis. Specifically, the proposed method estimates the grid state by analyzing the data collected from multiple sources. A low-rank function is learned to unveil the shared true data from observations, and the sparsity of data is applied to formulate bad data. Furthermore, an iterative optimization algorithm is proposed to solve the objective function. At last, extensive experiments on several IEEE bus systems verify the superiority of the proposed method.
With the widely deployment of information techniques in smart grid, it is quite important to automatically detect the bad data, e.g., malicious injection data and unfunctional sensor data, from daily observations. In this paper, we propose a novel approach for bad data detection in smart grid based on multi-view low-rank analysis. Specifically, the proposed method estimates the grid state by analyzing the data collected from multiple sources. A low-rank function is learned to unveil the shared true data from observations, and the sparsity of data is applied to formulate bad data. Furthermore, an iterative optimization algorithm is proposed to solve the objective function. At last, extensive experiments on several IEEE bus systems verify the superiority of the proposed method.
2019, 48(3): 366-373.
doi: 10.3969/j.issn.1001-0548.2019.03.010
Abstract:
For the symmetrical six-phase permanent magnet synchronous motor (PMSM) and three-phase PMSM series-connected system supplied by single inverter, the control current is transformed into the stationary coordinates composed of three mutually perpendicular subspaces through the decoupling transformation. The six-phase motor and the three-phase motor are controlled by current in the former two subspaces respectively. Furthermore, the mathematical model is transformed into the rotating coordinate system through rotation transformation. The vector control strategy is adopted to realize the independent control and decoupling operation. The experimental result shows that the symmetrical six-phase PMSM and three-phase PMSM series-connected system is able to achieve smooth operation and rapid dynamic response in which the status of one motor has no influence on the other.
For the symmetrical six-phase permanent magnet synchronous motor (PMSM) and three-phase PMSM series-connected system supplied by single inverter, the control current is transformed into the stationary coordinates composed of three mutually perpendicular subspaces through the decoupling transformation. The six-phase motor and the three-phase motor are controlled by current in the former two subspaces respectively. Furthermore, the mathematical model is transformed into the rotating coordinate system through rotation transformation. The vector control strategy is adopted to realize the independent control and decoupling operation. The experimental result shows that the symmetrical six-phase PMSM and three-phase PMSM series-connected system is able to achieve smooth operation and rapid dynamic response in which the status of one motor has no influence on the other.
2019, 48(3): 374-380.
doi: 10.3969/j.issn.1001-0548.2019.03.011
Abstract:
In order to analyze the influence of the Markova jumping parameters on the system, this paper deals with dynamic behavior analysis for a class of interval neural networks defined in complex number domain with Markova jumping parameters and time-varying delays. It is assumed that the activation functions defined in complex number domain satisfy Lipschitz condition. Firstly, the existence and uniqueness of the equilibrium point of the addressed system are studied by employing the M-matrix theory and the homeomorphism mapping theory. Then, the stochastic exponential robust stability of the equilibrium point is analyzed based on the idea of the vector Lyapunov function method. The presented stability analysis is the generalization of existing ones not only, but also easy to be verified in the practice applications. Finally, a numerical example with several simulation results is given to illustrate the feasibility of the obtained results in this paper.
In order to analyze the influence of the Markova jumping parameters on the system, this paper deals with dynamic behavior analysis for a class of interval neural networks defined in complex number domain with Markova jumping parameters and time-varying delays. It is assumed that the activation functions defined in complex number domain satisfy Lipschitz condition. Firstly, the existence and uniqueness of the equilibrium point of the addressed system are studied by employing the M-matrix theory and the homeomorphism mapping theory. Then, the stochastic exponential robust stability of the equilibrium point is analyzed based on the idea of the vector Lyapunov function method. The presented stability analysis is the generalization of existing ones not only, but also easy to be verified in the practice applications. Finally, a numerical example with several simulation results is given to illustrate the feasibility of the obtained results in this paper.
2019, 48(3): 381-387.
doi: 10.3969/j.issn.1001-0548.2019.03.012
Abstract:
The unsafe behavior of the driver is one of the important causes of many incidents. This paper presents a method to recognize unsafe driving behaviors based on the convolutional neural network. Firstly, the characteristics of the real-time image are extracted by the specific convolutional neural network, and then three kinds of behaviors are classified into two categories in parallel. The data set of unsafe driving behaviors in a real scene is established. The test on this dataset illustrates the efficiency and good generalization of the method. The experimental results show that the method achieves 99.85%, 99.62% and 98.68% accuracy for calling, smoking and unbelting in the data set of about 100 000 images, which is comparable to the results obtained by recent advanced Inception-v3 and Xception models.
The unsafe behavior of the driver is one of the important causes of many incidents. This paper presents a method to recognize unsafe driving behaviors based on the convolutional neural network. Firstly, the characteristics of the real-time image are extracted by the specific convolutional neural network, and then three kinds of behaviors are classified into two categories in parallel. The data set of unsafe driving behaviors in a real scene is established. The test on this dataset illustrates the efficiency and good generalization of the method. The experimental results show that the method achieves 99.85%, 99.62% and 98.68% accuracy for calling, smoking and unbelting in the data set of about 100 000 images, which is comparable to the results obtained by recent advanced Inception-v3 and Xception models.
2019, 48(3): 388-396.
doi: 10.3969/j.issn.1001-0548.2019.03.013
Abstract:
Network attacks that employ covert techniques to against security detections and achieve long-term latency and information theft have become major security issues in the current network. There are currently three challenges in this field. The strong concealment of the attack makes it difficult to detect, massive communication data in a high-speed network environment makes it difficult to build a detection model in a fine-grained manner, and the persistence and complexity of covert communication make the lack of tag data and increase the difficulty of model construction. Aiming at the above problems, based on the statistical analysis of campus network traffic, this paper describes and studies the hidden communication behavior based on covert conversation, and proposes a hidden communication behavior detection method. The original session flow is aggregated by parallelized session flow aggregation algorithm, and the covert communication behavior is characterized from the perspective of concentration trend and dispersion degree. The tag propagation algorithm is introduced to extend the tag data, and finally the multi-class detection model is constructed. The simulation results and the experiments in real network environment verify the detection effect of the method on the hidden communication behavior.
Network attacks that employ covert techniques to against security detections and achieve long-term latency and information theft have become major security issues in the current network. There are currently three challenges in this field. The strong concealment of the attack makes it difficult to detect, massive communication data in a high-speed network environment makes it difficult to build a detection model in a fine-grained manner, and the persistence and complexity of covert communication make the lack of tag data and increase the difficulty of model construction. Aiming at the above problems, based on the statistical analysis of campus network traffic, this paper describes and studies the hidden communication behavior based on covert conversation, and proposes a hidden communication behavior detection method. The original session flow is aggregated by parallelized session flow aggregation algorithm, and the covert communication behavior is characterized from the perspective of concentration trend and dispersion degree. The tag propagation algorithm is introduced to extend the tag data, and finally the multi-class detection model is constructed. The simulation results and the experiments in real network environment verify the detection effect of the method on the hidden communication behavior.
2019, 48(3): 397-401, 461.
doi: 10.3969/j.issn.1001-0548.2019.03.014
Abstract:
EPC Class-1 Generation-2 Version-2 (Gen2v2 for short) not only inherits the advantages of the original standard in terms of long reading range and large reading volume, but proposes a new security framework to enhance the security of radio frequency identification (RFID) system. Nowadays, the authentication protocol designed under the security framework has emerged as a hot topic. Based on the analysis of the existing Gen2v2-compliant authentication protocols, a new effective privacy preserving authentication protocol for Gen2v2 standard is proposed. The protocol is able not only to provide data confidentiality, anonymity and forward security, but also to withstand the tag impersonation attack, tracking attack, sniffing attack, replay attack and desynchronization attack. Compared with the existing works, the protocol has a lower communication and time overhead, and is more suitable for large-scale deployment.
EPC Class-1 Generation-2 Version-2 (Gen2v2 for short) not only inherits the advantages of the original standard in terms of long reading range and large reading volume, but proposes a new security framework to enhance the security of radio frequency identification (RFID) system. Nowadays, the authentication protocol designed under the security framework has emerged as a hot topic. Based on the analysis of the existing Gen2v2-compliant authentication protocols, a new effective privacy preserving authentication protocol for Gen2v2 standard is proposed. The protocol is able not only to provide data confidentiality, anonymity and forward security, but also to withstand the tag impersonation attack, tracking attack, sniffing attack, replay attack and desynchronization attack. Compared with the existing works, the protocol has a lower communication and time overhead, and is more suitable for large-scale deployment.
2019, 48(3): 402-408.
doi: 10.3969/j.issn.1001-0548.2019.03.015
Abstract:
Aiming at the problem that Mean Shift tracking algorithm cannot track well in complicated background, a Mean Shift algorithm based on fuzzy background weighting is proposed. It introduces a fuzzy membership function based on difference, which makes use of the difference between target model and background model in order to represents each pixel contribution to target exact description, and improves target description accuracy. At the same time, the original scale increment and decrement method is improved by using background information for adapting to target scale changing. Experimental results show that the proposed algorithm solves the problem of small-scale wandering and tracking hysteresis of the scale increment and decrement method to a certain extent, and improves the robustness of Mean Shift algorithm under complex background disturbances.
Aiming at the problem that Mean Shift tracking algorithm cannot track well in complicated background, a Mean Shift algorithm based on fuzzy background weighting is proposed. It introduces a fuzzy membership function based on difference, which makes use of the difference between target model and background model in order to represents each pixel contribution to target exact description, and improves target description accuracy. At the same time, the original scale increment and decrement method is improved by using background information for adapting to target scale changing. Experimental results show that the proposed algorithm solves the problem of small-scale wandering and tracking hysteresis of the scale increment and decrement method to a certain extent, and improves the robustness of Mean Shift algorithm under complex background disturbances.
2019, 48(3): 409-414.
doi: 10.3969/j.issn.1001-0548.2019.03.016
Abstract:
In this paper, a new type of principal component analysis (PCA) for electronic warfare information integration fusion scheme is presented. Based on DS theory of information fusion, the PCA method is used to collect data and reduce dimensions, and the basic trust distribution function is established for feature layer data to realize further based data fusion. The electronic warfare system information is intelligently diagnosed and mined to effectively achieve the fault detection and separation of electronic warfare system. Furthermore, the device status is evaluated and timely sent the control system through the big data mining, thus implementing the reasonable guidance and early warning and control for operational strategy in warfare procedure.
In this paper, a new type of principal component analysis (PCA) for electronic warfare information integration fusion scheme is presented. Based on DS theory of information fusion, the PCA method is used to collect data and reduce dimensions, and the basic trust distribution function is established for feature layer data to realize further based data fusion. The electronic warfare system information is intelligently diagnosed and mined to effectively achieve the fault detection and separation of electronic warfare system. Furthermore, the device status is evaluated and timely sent the control system through the big data mining, thus implementing the reasonable guidance and early warning and control for operational strategy in warfare procedure.
2019, 48(3): 415-419.
doi: 10.3969/j.issn.1001-0548.2019.03.017
Abstract:
Machine learning and template attack in the traditional side channel attack techniques have similar procedures, they all consist of two phases:learning and testing. Template attack can be considered as a classification technique for supervised learning, and there are many such classification algorithms in the machine learning field. In order to explore the application of machine learning algorithms in side channel attack, using the data movement operation in actual cryptographic chip as the attack target, the forecasting effect of some machine learning algorithms is investigated. These algorithms make use of power traces with known value of the moved data, then predict the value of the moved data for some new power traces. The results show that, when employing only one power trace in the testing stage, some machine learning algorithms have higher correctness rate than template attack.
Machine learning and template attack in the traditional side channel attack techniques have similar procedures, they all consist of two phases:learning and testing. Template attack can be considered as a classification technique for supervised learning, and there are many such classification algorithms in the machine learning field. In order to explore the application of machine learning algorithms in side channel attack, using the data movement operation in actual cryptographic chip as the attack target, the forecasting effect of some machine learning algorithms is investigated. These algorithms make use of power traces with known value of the moved data, then predict the value of the moved data for some new power traces. The results show that, when employing only one power trace in the testing stage, some machine learning algorithms have higher correctness rate than template attack.
2019, 48(3): 420-426.
doi: 10.3969/j.issn.1001-0548.2019.03.018
Abstract:
Collaborative filtering (CF) recommender system has been a most successful recommendation model in the past decade. However, the sparseness of user-item matrix and cold-tart problem still remain the challenges. The emergence of online social networking provides a great deal of social trust information for recommender systems, thus providing an opportunity to solve these problems. In this paper, based on matrix factorization collaborative filtering method, a model of integrating user trust information is proposed. This method uses trust information of users to amend the user latent factors and employs an auto-encoder to extract the initialization features of user and item latent feature vectors. And then a trust group detection algorithm is proposed for the trust relationship in the social network. Extensive experiments on real data sets show that the proposed model can not only effectively alleviate cold start, but also achieve better recommendation performance than the compared algorithms.
Collaborative filtering (CF) recommender system has been a most successful recommendation model in the past decade. However, the sparseness of user-item matrix and cold-tart problem still remain the challenges. The emergence of online social networking provides a great deal of social trust information for recommender systems, thus providing an opportunity to solve these problems. In this paper, based on matrix factorization collaborative filtering method, a model of integrating user trust information is proposed. This method uses trust information of users to amend the user latent factors and employs an auto-encoder to extract the initialization features of user and item latent feature vectors. And then a trust group detection algorithm is proposed for the trust relationship in the social network. Extensive experiments on real data sets show that the proposed model can not only effectively alleviate cold start, but also achieve better recommendation performance than the compared algorithms.
2019, 48(3): 427-431.
doi: 10.3969/j.issn.1001-0548.2019.03.019
Abstract:
A dynamic adjustment and active selection method for rough set decision making based on filtering function is proposed. Firstly, a sample filtering function is defined to determine the basis for sample selection or discarding; then, new samples are added in turn to determine the retention of samples according to the filtering function, and the decision-making tendency of existing samples is adjusted according to the threshold; finally, new sample library is established and attribute reduction is carried out. This method overcomes the problems of complex implementation process and large amount of calculation in traditional variable precision methods, and can effectively remove noise data and improve the robustness of the system. Experimental results show that this method can effectively compress data and improve the quality of sample analysis.
A dynamic adjustment and active selection method for rough set decision making based on filtering function is proposed. Firstly, a sample filtering function is defined to determine the basis for sample selection or discarding; then, new samples are added in turn to determine the retention of samples according to the filtering function, and the decision-making tendency of existing samples is adjusted according to the threshold; finally, new sample library is established and attribute reduction is carried out. This method overcomes the problems of complex implementation process and large amount of calculation in traditional variable precision methods, and can effectively remove noise data and improve the robustness of the system. Experimental results show that this method can effectively compress data and improve the quality of sample analysis.
2019, 48(3): 432-439.
doi: 10.3969/j.issn.1001-0548.2019.03.020
Abstract:
Link prediction can predict the missing links of complex networks, which promotes a better understanding of evolution mechanisms in real networks. Many similarity indices have been proposed based on a topology structure for link prediction. Local topological information, especially common neighbors, plays an important role in calculating the similarity between two endpoints. However, plenty of similarity indices ignore the effectiveness of common neighbors under different topology structures. Considering the local topological information around common neighbors, an effective common neighbor index is proposed. Firstly, we analyze the effectiveness of all neighbor links of common neighbors. Then, based on the local topology on both sides of two endpoints around common neighbors, the effectiveness of two sides of common neighbors is quantified separately. Finally, the similarity between two endpoints is described through the effect of common neighbors' effectiveness on bilateral resource allocation process. Empirical study on 15 real networks shows that the index proposed can achieve higher prediction accuracy, compared with 9 mainstream baselines.
Link prediction can predict the missing links of complex networks, which promotes a better understanding of evolution mechanisms in real networks. Many similarity indices have been proposed based on a topology structure for link prediction. Local topological information, especially common neighbors, plays an important role in calculating the similarity between two endpoints. However, plenty of similarity indices ignore the effectiveness of common neighbors under different topology structures. Considering the local topological information around common neighbors, an effective common neighbor index is proposed. Firstly, we analyze the effectiveness of all neighbor links of common neighbors. Then, based on the local topology on both sides of two endpoints around common neighbors, the effectiveness of two sides of common neighbors is quantified separately. Finally, the similarity between two endpoints is described through the effect of common neighbors' effectiveness on bilateral resource allocation process. Empirical study on 15 real networks shows that the index proposed can achieve higher prediction accuracy, compared with 9 mainstream baselines.
2019, 48(3): 440-448.
doi: 10.3969/j.issn.1001-0548.2019.03.021
Abstract:
Community detection is of great significance for exploring the structural characteristics of complex networks while the performance of community detection algorithm makes important influence on the detection results. At present, the benchmark networks that are used to measure the performance of community detection algorithm mainly include artificial synthetic network and real-world network. Synthetic network has become the main method to measure the performance of the algorithm since the real-world network usually lacks information of known community structure. However, it is found that the microscopic characteristics of the network is unadjusted, which is different from the real-world network, the discrimination of the detection algorithm is not high, and it is inability to change the local network structure. In order to improve the performance of artificial synthetic network, a benchmark network construction algorithm on null-model is proposed. Firstly, an algorithm of null model that can maintain the mesoscale characteristics is built to improve the flexibility of network micro-feature adjustment and make it closer to the real-world network structural characteristics. Secondly, the null model of adjusting strengthen and weakness for community structure is designed for improving the evaluation accuracy of network community testing. Finally, a method based on null model is constructed so as to make some adjustments of the local topological structure for measuring the importance of the change with local community structure characteristics to the whole network structure and the performance on detection algorithm. Experimental results show that the algorithm in view of null model can effectively improve the diversity and flexibility of the benchmark network, thus making the network be more similar with the features of real-world network and meeting the demand for performance improvement of community detection algorithm.
Community detection is of great significance for exploring the structural characteristics of complex networks while the performance of community detection algorithm makes important influence on the detection results. At present, the benchmark networks that are used to measure the performance of community detection algorithm mainly include artificial synthetic network and real-world network. Synthetic network has become the main method to measure the performance of the algorithm since the real-world network usually lacks information of known community structure. However, it is found that the microscopic characteristics of the network is unadjusted, which is different from the real-world network, the discrimination of the detection algorithm is not high, and it is inability to change the local network structure. In order to improve the performance of artificial synthetic network, a benchmark network construction algorithm on null-model is proposed. Firstly, an algorithm of null model that can maintain the mesoscale characteristics is built to improve the flexibility of network micro-feature adjustment and make it closer to the real-world network structural characteristics. Secondly, the null model of adjusting strengthen and weakness for community structure is designed for improving the evaluation accuracy of network community testing. Finally, a method based on null model is constructed so as to make some adjustments of the local topological structure for measuring the importance of the change with local community structure characteristics to the whole network structure and the performance on detection algorithm. Experimental results show that the algorithm in view of null model can effectively improve the diversity and flexibility of the benchmark network, thus making the network be more similar with the features of real-world network and meeting the demand for performance improvement of community detection algorithm.
2019, 48(3): 449-455.
doi: 10.3969/j.issn.1001-0548.2019.03.022
Abstract:
In recent years, various recommendation methods have been proposed by referring to processes originated in statistical physics, among them the diffusion-based method is an important branch of study. However, these methods were proposed solely based on rating metrics, while the trust relations among users are always ignored. In this paper, we propose a novel information filtering algorithm by introducing users' social trust relationships into the original diffusion-based method based on the resource-allocation process. Specifically, a tunable parameter is used to scale the resources received by trusted users in the networked resource redistribution process. The objects collected by trusted users will receive more resources. Extensive experiments on the two real-world rating and trust datasets, Epinions and FriendFeed, suggest that the proposed algorithm has better performance than benchmark algorithms in terms of accuracy, diversity, and novelty in the recommendation.
In recent years, various recommendation methods have been proposed by referring to processes originated in statistical physics, among them the diffusion-based method is an important branch of study. However, these methods were proposed solely based on rating metrics, while the trust relations among users are always ignored. In this paper, we propose a novel information filtering algorithm by introducing users' social trust relationships into the original diffusion-based method based on the resource-allocation process. Specifically, a tunable parameter is used to scale the resources received by trusted users in the networked resource redistribution process. The objects collected by trusted users will receive more resources. Extensive experiments on the two real-world rating and trust datasets, Epinions and FriendFeed, suggest that the proposed algorithm has better performance than benchmark algorithms in terms of accuracy, diversity, and novelty in the recommendation.
2019, 48(3): 456-461.
doi: 10.3969/j.issn.1001-0548.2019.03.023
Abstract:
Based on the nearest neighbor information of the node, the resilience of complex systems can be measured by using prediction model for complex system resilience through mapping multidimensional equation into one-dimensional equation. However, this model does not introduce the second-order neighbor information of the node. In this paper, we present a prediction model of the resilience of complex systems by considering the second-order neighbor information. Then using the Barabási-Albert (BA) scale free network and Watts-Strogatz (WS) small world network, we investigate the effect of improved model and explore the impact of improved model with different network structures. The experiment results show that for the BA scale free network and WS small-world network with different average degree of network, the improved model considering with the second-order neighbor information can predict the resilience of complex systems more accurately. When the average degrees of the BA scale free network and WS small-world network are 2, the accuracies of system resilience measurement are increased by 79.89% and 59.53%. For the same kind of networks the smaller the average degree of networks is, the more accurate the prediction of the improved model is. Then we also find that when the size and average degree of networks are the same, the effect of improved model is better for BA scale-free network than WS small-word network. Our researches provide theoretical support and research method for measuring resilience of complex networks and designing resilient systems.
Based on the nearest neighbor information of the node, the resilience of complex systems can be measured by using prediction model for complex system resilience through mapping multidimensional equation into one-dimensional equation. However, this model does not introduce the second-order neighbor information of the node. In this paper, we present a prediction model of the resilience of complex systems by considering the second-order neighbor information. Then using the Barabási-Albert (BA) scale free network and Watts-Strogatz (WS) small world network, we investigate the effect of improved model and explore the impact of improved model with different network structures. The experiment results show that for the BA scale free network and WS small-world network with different average degree of network, the improved model considering with the second-order neighbor information can predict the resilience of complex systems more accurately. When the average degrees of the BA scale free network and WS small-world network are 2, the accuracies of system resilience measurement are increased by 79.89% and 59.53%. For the same kind of networks the smaller the average degree of networks is, the more accurate the prediction of the improved model is. Then we also find that when the size and average degree of networks are the same, the effect of improved model is better for BA scale-free network than WS small-word network. Our researches provide theoretical support and research method for measuring resilience of complex networks and designing resilient systems.
2019, 48(3): 462-468.
doi: 10.3969/j.issn.1001-0548.2019.03.024
Abstract:
According to the inaccurate description of various factors in the traditional information propagation rate model and the high error of simulation results, an information propagation rate model considering propagation speed (PPS) is proposed. First, the model selects the Digg social platform and analyzes the data set; Secondly, a large amount of data is used to improve the calculation method of intrinsic growth rate and user carrying capacity of the traditional model, then the PPS model is obtained according to the different news voting quantity, and finally the simulation analysis is carried out for different coverage stories. The simulation results show that the news on this platform has experienced the growth period and reached the stable period, and the speed of the news is the fastest after the front page is shown. Based on the traditional model and the algorithm, the proposed PPS model has a great improvement in accuracy, which proves that the model is reasonable and effective in analyzing the information transmission rate of the social platform.
According to the inaccurate description of various factors in the traditional information propagation rate model and the high error of simulation results, an information propagation rate model considering propagation speed (PPS) is proposed. First, the model selects the Digg social platform and analyzes the data set; Secondly, a large amount of data is used to improve the calculation method of intrinsic growth rate and user carrying capacity of the traditional model, then the PPS model is obtained according to the different news voting quantity, and finally the simulation analysis is carried out for different coverage stories. The simulation results show that the news on this platform has experienced the growth period and reached the stable period, and the speed of the news is the fastest after the front page is shown. Based on the traditional model and the algorithm, the proposed PPS model has a great improvement in accuracy, which proves that the model is reasonable and effective in analyzing the information transmission rate of the social platform.
2019, 48(3): 469-474.
doi: 10.3969/j.issn.1001-0548.2019.03.025
Abstract:
In order to reveal the electroencephalogram (EEG) differences between truth and lying, the event related synchronization (ERS) is deeply studied for spontaneous EEG during the lying detection experiment. The ERS corresponding to delta and theta rhythm of EEG is analyzed in detail. The results indicate that ERS corresponding to delta and theta rhythm can reflect the deception-related psychological states. The delta synchronization is strongly related with P300. The theta synchronization effect overlaps with the ERPs effect, but ERPs and the synchronization effects have different scalp distribution character. The results show that the theta synchronization effect is strongest in Fz electrode and the ERPs effect is maximal on Pz electrode. This evidence implies that ERS corresponding to theta rhythm and P300 might reflect the deception-related cognitive psychological process from different aspects.
In order to reveal the electroencephalogram (EEG) differences between truth and lying, the event related synchronization (ERS) is deeply studied for spontaneous EEG during the lying detection experiment. The ERS corresponding to delta and theta rhythm of EEG is analyzed in detail. The results indicate that ERS corresponding to delta and theta rhythm can reflect the deception-related psychological states. The delta synchronization is strongly related with P300. The theta synchronization effect overlaps with the ERPs effect, but ERPs and the synchronization effects have different scalp distribution character. The results show that the theta synchronization effect is strongest in Fz electrode and the ERPs effect is maximal on Pz electrode. This evidence implies that ERS corresponding to theta rhythm and P300 might reflect the deception-related cognitive psychological process from different aspects.
2019, 48(3): 475-480.
doi: 10.3969/j.issn.1001-0548.2019.03.026
Abstract:
Visual perceptual learning with aberration corrected by adaptive optics visual stimulator system (AOVS) is applied to human eyes. Four subjects with anisometropia and four subjects with strabismus are trained by this way. Statistical results show that based on the same method, the contrast sensitivity function (CSF) of four subjects with anisometropia are improved significantly than that of four subjects with strabismus. The comparison of visual acuity (VA) between two groups is performed before and after the training. Experiment results prove that the proposed method is effective for subjects with anisometropia but not effective for subjects with strabismus.
Visual perceptual learning with aberration corrected by adaptive optics visual stimulator system (AOVS) is applied to human eyes. Four subjects with anisometropia and four subjects with strabismus are trained by this way. Statistical results show that based on the same method, the contrast sensitivity function (CSF) of four subjects with anisometropia are improved significantly than that of four subjects with strabismus. The comparison of visual acuity (VA) between two groups is performed before and after the training. Experiment results prove that the proposed method is effective for subjects with anisometropia but not effective for subjects with strabismus.
