Loading...
  • 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

Current Issue

    • Server power optimization based on  predictive thermal dissipation
      LIN Kai-zhi, ZONG Yan-yan, ZHANG Yan-nan,
      2020, 42(08): 1331-1338. doi:
      Abstract ( 203 )   PDF (1575KB) ( 167 )     
      With the rapid development of the Internet industry and the advent of the 5G era, the demand for high-performance servers and storage devices is increasing. However, at the same time, the requirement for power consumption optimization is also increasing. A control strategy based on predictive thermal dissipation for server power optimization is proposed. Firstly, the maximum temperature of the device is obtained. It is set as a reference point after it is compared with the temperature threshold. The temperature trend is predicted by evaluating the current change at the reference point. Then, a corresponding control signal is sent to adjust the fan speed in order to achieve the purpose of power optimization. Finally, an experimental system was built up. For the power optimization problem of the system, three sets of experiments (the same duty cycle in different periods, different duty cycles in the same period, and different temperatures) are designed to verify the proposed control strategy. The experimental results show that the control strategy can effectively reduce the devices’ power consumption.
      Load balancing optimization of consistent hashing  microservice based on dynamic weight
      ZHANG Kai-qi, LIU Xiao-yan, WANG Xin, JI Chun-shan, YAN Xin
      2020, 42(08): 1339-1344. doi:
      Abstract ( 487 )   PDF (543KB) ( 239 )     
      With the development of Internet technology, the load capacity of Internet server clusters is facing unprecedented challenges, and it is particularly important to implement a reasonable load balancing strategy. In order to achieve the best efficiency for load balancing, a consistent hash algorithm is adopted to distribute the loads of the load balancing system into a cluster. This paper analyzes the characteristics of load balancing in a cluster of servers based on microservice architecture, and proposes a method for designing and segmenting a consistent hash ring based on virtual nodes and a dynamic weight-based allocation strategy. Based on the consistent hash algorithm, load transfer between service clusters is achieved to solve the problem of load imbalance between services caused by increased service load in microservice clusters. It can prevent some services from crashing due to excessive load pressure. The experiments show that the improved load balancing strategy only has 31% load imbalance probability of the traditional consistency algorithm, and the dynamic allocation strategy has good load balancing performance, and can effectively solve the load balancing problem of microservices distributed architecture.
      A service-oriented TT&C service  bus system of survey ship
      LI Yong-gang, LI Xiang-ming, WU Yun, YANG Hai-min, MAO Wen
      2020, 42(08): 1345-1351. doi:
      Abstract ( 129 )   PDF (1138KB) ( 119 )     
      The current TT&C software system of survey ship needs to complete the processing of data such as telemetry, remote control, tracking measurement, simulation of various aircrafts. The system is relatively complex and difficult to expand and maintain. SOA (Service Oriented Architecture) is a component model, which encapsulates different functional units of a complex system as services and rea- lizes the invocation of services through the interface, greatly reducing the complexity of the system and facilitating the expansion of system functions. This paper analyzes the TT&C software system of survey ship and designs the architecture of service-oriented TT&C service bus system of survey ship. The architecture supports two service scheduling modes: “service scheduling rules” and “service scheduling system”. The server, client and service scheduling system are implemented. The test results show that the system can guarantee the performance of service scheduling under two modes, which is of great significance to improve the efficiency of the TT&C software system of survey ship. 

      A connectivity restoration algorithm based on Steiner tree and Tyson polygon
      WANG Mao-qiu, ZHANG Jiang, ZHANG Jing,
      2020, 42(08): 1352-1358. doi:
      Abstract ( 174 )   PDF (808KB) ( 132 )     
      Aiming at the problem that wireless sensor networks are susceptible to severe environmental damage and the energy consumption of key nodes is fast after connectivity restoration, which leads to network disconnection, this paper proposes a network connectivity restoration (CRAST) algorithm based on Steiner tree and Tyson polygon. Firstly, the algorithm abstracts the divided node partitions into discrete points, enumerates all non-degenerate quads in the discrete point area, and then uses the quadrangular Steiner tree structure to deploy relay nodes to these non-degenerate quads so as to achieve connectivity restoration. Secondly, the algorithm constructs a Delaunay triangle network with key nodes, and uses the Delaunay triangle network to construct the Tyson polygon topology of the entire wireless sensor network. Finally, the algorithm deploys movable backup relay nodes at all vertices of the Tyson polygon, and moves the backup relay node to replace the damaged key node when the key node is damaged. The algorithm enables the sensor network to achieve connectivity recovery at the least cost, and has strong efficiency and robustness.

      An optimal filtering and clipping technique and a neural network based realization scheme for cubic metric reduction in OFDM system
      YUAN Tian, ZHU Hong-liang, ZHOU Juan, ZHU Xiao-dong
      2020, 42(08): 1359-1366. doi:
      Abstract ( 205 )   PDF (809KB) ( 147 )     
      One of the main drawbacks of Orthogonal Frequency Division Multiplexing (OFDM) signals is the large signal envelope fluctuation. Peak to Average Power Ratio (PAPR) is a commonly used metric for quantifying the envelope fluctuations of OFDM signals. However, recent researches have shown that Cubic Metric (CM) is a more accurate metric when it is used to quantify the envelope fluc- tuation. Clipping and filtering technique can be employed to reduce the CM. The filtering operation in traditional clipping and filtering technique cannot lead the signal to the optimal performance. Therefore, an optimal clipping and filtering algorithm for CM reduction is proposed. The key idea is to consider the impact of filtering operation on in-band and out-of-band components of signals and model the filter design as an optimization problem. The problem is solved to obtain an optimal filter, which is combined with clipping to reduce CM efficiently. Due to the high complexity of solving the optimization problem, a deep neural network based realization scheme of the optimal clipping and filter algorithm is further proposed. Simulation results show that both the proposed algorithm and corresponding neural network scheme have close performance, but the latter has much lower complexity. Compared with some existing algorithms, the proposed algorithm and scheme exhibit better performance.
      Probabilistic extension of Mediator
      XUE Xiao-yong, SUN Meng
      2020, 42(08): 1367-1373. doi:
      Abstract ( 283 )   PDF (517KB) ( 169 )     
      Mediator is a component-based formal modeling language. With its hierarchical and modular structure, it can easily model complex systems. It takes automaton as its underlying unit. Automatons are connected to form a system, which is a high-level architecture and easy to use while formalizing the model. In order to make Mediator have stronger expressive ability and formally model the system with probabilistic behavior, Mediator is extended in terms of probability, and the semantics based on Markov decision process are given to the extended language. In addition, an automatic PRISM code generation method is introduced, which can use PRISM to verify the relevant properties of Mediator models.

      A coverage table generation algorithm based on improved whale optimization
      LIU Xiang-ting, CAO Xiao-peng
      2020, 42(08): 1374-1382. doi:
      Abstract ( 138 )   PDF (544KB) ( 100 )     
      In order to improve the efficiency of coverage table generation in combination test, an improved whale optimization algorithm based on the discreteness of coverage table generation is proposed. Firstly, it uses the idea of encoding transformation to change the continuous motion mode of individual whale to the discrete mode of coverage table generation. Secondly, an iterative evolution operator is added to further improve the search ability in the development and search phase of the algorithm. Finally, aiming at the limitation problem of the algorithm itself in coverage table generation, it introduces the average hamming distance to jump out of the local optimum. Constraint handing mechanism is added by constraint solver and penalty function method, so as to improve the practical application ability of the algorithm. Experimental results show that the proposed algorithm has better advantages than he other existing algorithms in the scale of coverage table generation.

      Formal design and generation of a family of multiple sequences alignment algorithms
      ZHANG Xu-chu, SHI Hai-he
      2020, 42(08): 1383-1392. doi:
      Abstract ( 154 )   PDF (1015KB) ( 150 )     
      Multiple sequence alignment is an important part of bioinformatics research, which is the basis to solve the problems of species evolution relationship and genome sequence analysis. Multiple sequence alignment algorithms have high specificity and different algorithms are suitable for different research environments. At present, the commonly used multiple sequence alignment software is based on the assembly of multiple sub-algorithms under the guidance of bioinformatics theory. However, the existing research mainly focuses on the optimization of specific steps of specific algorithms, and the lack of algorithm framework research with high abstraction of domain level leads to the complexity and redundancy of multiple sequence alignment algorithms. According to the idea of generative programming and software reuse, the features of 
      multiple sequence alignment algorithm (MSAA) are analyzed and modeled, and the corresponding generic algorithm components are designed and the interaction among these components is described. Further, PAR platform is used to formally build up the MSAA component library. The work improves the reliability and assembly flexibility of assembling algorithms and facilitates the maintenance and optimization of researchers.






      A survey of video abnormal event detection
      WANG Si-qi, HU Jing-tao, YU Guang, ZHU En, CAI Zhi-ping
      2020, 42(08): 1393-1405. doi:
      Abstract ( 593 )   PDF (987KB) ( 406 )     
      Video anomaly detection is one of the most significant research tasks in computer vision area. It aims to intelligently identify the events that do not conform to expected behavior based on pattern recognition and computer vision methods. Video anomaly detection is widely applied and there is an enormous potential demand in modern society. Meanwhile, inspired by the successful achievements in various area of emerging deep learning technologies, more and more newly-emerged methods are conducted on video anomaly detection problem. Firstly, we retrospect the definition and main challenges of vi- deo anomaly detection. Secondly, we introduce the mainstream video anomaly detection methods from three primary technical steps (video event extraction, video event representation, video event modeling and detection) of video anomaly detection, and conclude their advantages as well as drawbacks respectively. Finally, we introduce the benchmark datasets and evaluation metrics of video anomaly detection, compare the performance of mainstream methods and give conclusions and prospects.

      A robust target tracking algorithm based on VGG network
      XU Liang, ZHANG Jiang, ZHANG Jing, YANG Ya-qi
      2020, 42(08): 1406-1413. doi:
      Abstract ( 206 )   PDF (1475KB) ( 207 )     
      In the traditional target tracking algorithm, when the target is disturbed by various factors such as occlusion and light intensity changes, the correlation filter template updates incorrectly and the error accumulates frame by frame, eventually causing the target tracking failure. Therefore, this paper proposes a robust object tracking algorithm based on VGG network. Firstly, the VGG network is used to extract the average feature map of the local context area image to establish a correlation filter template in the first frame of the input image. Secondly, the VGG network is used to extract the average feature map and the affine transformation average feature map of the local context area image in the subsequent frame of the input image. Thirdly, combining the kernel correlation filter tracking algorithm, the target position and the final target position are adaptively determined. Finally, the algorithm adaptively updates the final average feature map and the final correlation filter template in the current frame of the input image. Experimental results show that the proposed algorithm still has high target tracking accuracy and robustness when the target is disturbed by various factors such as occlusion and light intensity changes.
      Prediction of breast cancer based on C-AdaBoost model
      LI Yong, CHEN Si-xuan, JIA Hai, WANG Xia
      2020, 42(08): 1414-1422. doi:
      Abstract ( 187 )   PDF (921KB) ( 138 )     
      Machine learning and deep learning techniques can be used to solve many problems in me- dical classification prediction. Among them, some have higher prediction accuracy, but the others have limited accuracy. This paper proposes an ensemble learning algorithm based on C-AdaBoost model to predict breast cancer diseases. Stepwise regression is used to re-select existing features. The C-AdaBoost model is combined to make the prediction better. A large number of experiments show that 1) the optimal combination of features, that determines whether breast cancer recurs and whether breast cancer is benign, are found, and 2) the proposed ensemble learning algorithm based on C-AdaBoost improves the prediction accuracy by at most 19.5% in comparison to the machine learning classifiers such as SVM, Naive Bayes, RandomForest and traditional ensemble learning models, which can better help doctors make clinical decisions.

      Trajectory control of ball and plate system based on LM_RBF-PID
      HUANG Wen-jie, XIANG Feng-hong, MAO Jian-lin
      2020, 42(08): 1423-1429. doi:
      Abstract ( 139 )   PDF (924KB) ( 106 )     
      Ball and plate system is a typical two-input two-output system. Aiming at the problems of slow response and serious shocks in RBF-PID control algorithm. The Lagrange equation is used to model the ball and plate system under the condition of ignoring the interference factors. Radical Basis Function (RBF) neural network is used to identify the discrete model of ball and plate system on line and realize adaptive control. On this basis, the Levenberg-Marquardt (LM) algorithm replaces the gradient descent method to set the control parameters. The LM_RBF-PID controller is designed and compared with the RBF-PID controller in terms of step signal response and square wave signal response. Finally, a square trajectory tracking experiment is completed in the ball and plate system. The experimental results show that proposed algorithm improves trajectory tracking control accuracy, and it can ensure the stability and convergence of the tracking control in the ball and plate system.
      Weighted hesitant triangular fuzzy distance measurement and its application in group decision making
      MA Hui, WEI Li-li
      2020, 42(08): 1430-1439. doi:
      Abstract ( 137 )   PDF (472KB) ( 97 )     
      Considering the different importance of different elements in hesitant triangular fuzzy elements as membership, the concepts of weighted hesitant triangular fuzzy elements and weighted hesitant triangular fuzzy sets are proposed, the group decision problem with decision value as weighted hesitant triangular fuzzy element is studied. Firstly, the formula of weighted hesitant triangular fuzzy distance is given. Secondly, in order to facilitate calculation and do not change the importance of triangular fuzzy numbers as membership, a method of adding elements to weighted hesitant triangular fuzzy elements is proposed. Finally, a group decision-making method based on weighted hesitant triangular fuzzy distance measurement is proposed and applied to group decision-making under the weighted hesitant triangular fuzzy environment. Numerical examples show that the weighted hesitant triangular fuzzy distance metric has rationality and feasibility in group decision making. 
      A semi-supervised outlier detection model based on autoencoder and integrated learning
      XIA Huo-song, SUN Ze-lin
      2020, 42(08): 1440-1447. doi:
      Abstract ( 281 )   PDF (933KB) ( 197 )     
      Outlier detection is an important data mining method, which is used to preprocess data and mine heterogeneous data information. In recent years, due to the problem of dimension disaster, it is very difficult to detect the high-dimensional outlier data. Aiming at the above problems, a semi- supervised outlier detection model based on autoencoder and integrated learning is proposed. Firstly, autoencoder is used to reduce the dimension and increase the outlier degree of the outlier data. Secondly, considering that Iforest, lof and k-means algorithms are sensitive to different outlier types, they are fused in the AdaBoost boosting framework to improve the accuracy of outlier detection. The results show that, compared with the current mainstream outlier detection methods, the proposal significantly improves the accuracy of the model.
      A binary-search enhanced Karnik-Mendel algorithm for type-reduction of general type-2 fuzzy logic system
      CHEN Yang, WANG Tao
      2020, 42(08): 1448-1453. doi:
      Abstract ( 213 )   PDF (621KB) ( 136 )     
      General type-2 fuzzy logic systems have drawn great attentions in recent years, and type-reduction is still the kernel block of the systems. The Enhanced Karnik-Mendel (EKM) algorithm is the most popular type-reduction algorithm. According to the alpha-planes representation theory of general type-2 fuzzy sets, this paper proposes a binary-search enhanced Karnik-Mendel (BEKM) algorithm to perform the centroid type-reduction of general type-2 fuzzy logic systems. In case of choosing the same sampling rate of primary variables, two computer simulation examples show that, compared with the EKM algorithm, the BEKM algorithm has higher computational efficiency without losing the calculation accuracy.

      A dynamic multi-objective evolutionary algorithm based on Pareto solution set segmentation prediction strategy
      MA Yong-jie, CHEN Man-li, CHEN Min
      2020, 42(08): 1454-1462. doi:
      Abstract ( 131 )   PDF (1613KB) ( 111 )     
      Aimed at the problems of slow convergence and difficulty in maintaining diversity, a dynamic multi-objective evolutionary algorithm based on Pareto solution set segmentation prediction is proposed. When the environmental change is detected, the Pareto optimal solution obtained from the evolution at the previous moment is sorted according to a sub-objective function  and divided into three segments according to the size of the sub-objective, then the moving direction of the center point of each Pareto solution set is calculated. Each Pareto subset is systematically sampled to obtain the feature points of the Pareto frontier surface, and the linear model is used to predict the next generation population. According to the difficulty of the optimization problem, adaptive random populations are generated around the predicted population to increase the diversity of the population. 
      Experiments on the three types of standard test functions show that the algorithm can effectively solve the dynamic multi-objective optimization problem.

      Local generalized multi-granulation rough set
      WANG Hong, LI Min-ying
      2020, 42(08): 1463-1471. doi:
      Abstract ( 146 )   PDF (490KB) ( 120 )     
      The target concept of multi-granulation rough sets is a kind of granular structure approximation induced by multiple binary relations, which is a valuable direction in the field of rough sets and has been widely used in practice. However, there are a lot of unlabeled data when the data set is large, and calculating the approximation of the target concept requires to calculate the equivalent class of all objects, which takes a lot of time to describe the approximation of the target concept as well as the complicated calculation process. Therefore, a local generalized multi-granulation rough set model is proposed. Firstly, the lower and upper approximations are defined by introducing characteristic functions. Secondly, a matrix method is proposed to solve the lower approximation and the upper approximation of the local generalized multi granularity rough set, and their properties are further studied. Finally, an example is given to verify the effectiveness of the proposed model, the algorithm to find the lower approximation of the local generalized multi granularity rough set is given. Besides, the model can make full use of the data information in the target concept to process data, which saves a lot of calculating time.

      A multi-objective particle swarm optimization algorithm with star structure to solve the multi-modal multi-objective problem
      GAO Hai-jun, PAN Da-zhi
      2020, 42(08): 1472-1481. doi:
      Abstract ( 179 )   PDF (1958KB) ( 127 )     
      Firstly, according to the particle structure information in the multi-objective particle swarm optimization algorithm, using non-dominated solution sets to construct the topological structure between individual particle neighborhoods, a star-structured multi-objective particle swarm optimization algorithm is proposed for solving multi-modal multi-objective problems. Secondly, in view of the difficulty of selecting the global optimal individual in the multi-objective particle swarm, an evaluation method for the uniformity of the distribution of non-dominated solution sets is proposed. The evaluation result determines the global optimal individual corresponding to the current particle. Finally, combining two methods, a star topology multi-objective particle swarm optimization algorithm with uniform calculation method is proposed. The test function analyzes the convergence of the algorithm and shows that the improved algorithm converges faster than the original algorithm. Experimental results show that the algorithm can take into account the distribution of the problem object space and decision space, and effectively solve the multi-modal multi-objective problem.

      An EEG signal recognition algorithm based on sample entropy and BP neural network
      SHEN Xiao-yan, WANG Xue-mei, WANG Yan
      2020, 42(08): 1482-1488. doi:
      Abstract ( 143 )   PDF (918KB) ( 188 )     
      Brain-Computer Interface (BCI) is an emerging technology for communication between human brain and external devices. The traditional feature extraction method based on time-frequency cha- racteristics cannot reflect the nonlinear characteristics of EEG signals. In order to further improve the accuracy of classification, the pretreatment method of wavelet threshold denoising is firstly used to improve the signal-to-noise ratio of EEG signals. Then, the feature extraction of the three kinds of imaginary motion EEG signals is carried out by the parameter-sample entropy of nonlinear dynamics, and the nonlinear features of EEG signals are preserved. Among them, the research of Motor-Imagery (MI) EEG has always been the focus of BCI that is a high-speed development field. This paper also studies three classifiers including support vector machine, LVQ neural network and BP neural network. The experimental results show that BP neural network has higher recognition rate for classification and recognition of EEG signals.
      A label propagation algorithm combining eigenvector centrality and label entropy
      PAN Shu-can, XU Qing-lin
      2020, 42(08): 1489-1499. doi:
      Abstract ( 147 )   PDF (757KB) ( 110 )     
      Overlapping community structure mining aims to discover overlapping parts between multiple independent communities in a complex network, and it has a wide range of applications in social, transportation, public opinion, and even anti-terrorism fields. However, the current overlapping community mining algorithm based on label propagation shows strong randomness in networks with fuzzy community structure, resulting in low accuracy. Aiming at the problem of uncertainty and low accuracy caused by fuzzy boundaries of overlapping communities, a label propagation algorithm combining eigenvector centrality and label entropy (ECLE-LPA) is proposed. ECLE-LPA utilizes the K-nuclear iteration factor and the eigenvector centrality of the node to calculate the node influence and initialize the node label. In the propagation process, the label entropy and the closeness of the nodes are calculated to update the label list and the label membership, so as to overcome the recognition problem of fuzzy boundaries of overlapping communities. The experimental results show that: in real networks such as Les, Polbooks, Football, Polblogs, Netscience, etc., the EQ value of ECLE-LPA algorithm is generally increased by 1%~3% compared with the contrast algorithm. In the artificial network, the NMI value of ECLE-LPA is more than 10% higher than that of the contrast algorithm.