Current Issue

• 2018, No. 06　Published:25 June 2018 Last issue    Next issue
  Columns in this issue:

• A function-level compiler optimization parameter

  selection method based on supervised learning model

  LIU Hui1,2，ZHAO Rongcai1，WANG Qi1

  2018, 40(06): 957-968. doi:

  Abstract ( 131 )   PDF (989KB) ( 264 )   　　 Review attachment

  The machine learning based iterative compilation can effectively predict new programs’ best optimization parameter combination. The existing methods suffer some problems, such as low search efficiency of optimization parameter combination, inappropriate representation of programs features and unsatisfactory prediction accuracy in model training process. The machine learning based iterative compilation is a hotspot research in the field of iterative compilation, and its challenge lies in choosing learning algorithms, optimizing parameter search and program features representation. Based on the supervised learning technique, we propose an optimization parameter predictive method, called SLOPS. We search the optimal parameter space by constraining the multiobjective PSO algorithm, and find the best optimization parameters of the sample function. Then we extract the features of the function through dynamic and static program feature representation techniques. Finally, a supervised learning model is constructed by the KNN and the Softmax regression based on the samples composed of program features and optimization parameters, which is used to predict the optimization parameters of new programs. Experimental results show that the SLOPS can achieve better prediction performance on NPB benchmarks and scientific programs.

   

  Design and implement of an eMMC

  controller in HS400 mode based on FPGA

  ZHANG Yu1,2,CHEN Wei1,WU Lizhou1,2,XIAO Nong1,2

  2018, 40(06): 969-976. doi:

  Abstract ( 653 )   PDF (1069KB) ( 457 )   　　 Review attachment

  We introduce the working principle of eMMC chip and its HS400 high speed data transmission mode, the design scheme of an eMMC controller, and the result of hardware verification. An eMMC controller that uses the DDR transmission mode for data transmission is implemented at a working frequency of 200 MHz, and the CRC verification module is employed to achieve the transmission data CRC checksum to enhance system reliability. The experimental platform adopts the motherboard/subboard architecture, and the eMMC controller is ultimately implemented on the Xilinx Zynq7000 FPGA Zedboard development board to realize communication through the FMC interface and eMMC chip board. Simulation and boardlevel tests show that data read and write operations in the HS400 mode can achieve a data transfer rate of up to 400 MB/s, which can effectively improve the access performance of the device during the actual eMMC development.

   

  A P2P network trust model for selecting recommended nodes

  MA Manfu1,2,HE Chunling1,2

  2018, 40(06): 977-983. doi:

  Abstract ( 100 )   PDF (685KB) ( 141 )   　　 Review attachment

  Due to the openness and anonymity of PeertoPeer (P2P), the types of interactions between nodes are complex and diverse, and they have small repetitive interactions, which makes it difficult to define the trust relationship between nodes. Based on selected nodes, we propose a P2P network trust model, and obtain comprehensive trust value through direct trust and the recommended trust for each node. Recommendation nodes whose trust value is closest to the interactive nodes are selected, that is to say, nodes with the smallest trust difference are selected. We also propose a joint punishment mechanism and an incentive mechanism to prevent gang cheating of the conspiracy group. Simulation experiments show that the trust model is not only capable of resisting attacks of malicious nodes, but also has a high selfadaptability.

   

  A deep learning based program

  performance analysis framework

  FENG Yunlong，LIU Yong，HE Wangquan

  2018, 40(06): 984-991. doi:

  Abstract ( 110 )   PDF (930KB) ( 275 )   　　 Review attachment

  The increasing complexity of high performance computer system architecture and inadequate intelligence of the existing performance analysis tools, lead to the increasing cost for performance analysis and optimization of high performance computing applications. Thanks to the deep learning technology, significant progress in the field of artificial intelligence has been made, and it also presents an opportunity for intelligent program performance analysis tools. We propose an intelligent program performance analysis framework, which abstracts program performance analysis into a classification problem via machine learning technology. The performance data, the input of the analysis framework, which is collected via the performance monitoring unit, is standardized in advance; and the categories of performance problems, the output of the analysis framework, are determined by the cluster evaluation and the actual meaning of every cluster. Sparse coding is used to automatically learn the features of performance data and a performance problem classification model is built. Finally, we implement the prototype of the framework on the Sunway TaihuLight system. Experimental results show that this framework can intuitively guide programmers to grasp the most prominent performance bottleneck problem in current applications, effectively improve the performance analysis efficiency, and reduce the cost of code tuning.

   

  A distributed storage and query method of

  remote sensing data based on HBase

  JING Weipeng，TIAN Dongxue

  2018, 40(06): 992-998. doi:

  Abstract ( 163 )   PDF (702KB) ( 202 )   　　 Review attachment

  The storage and query of remote sensing image is an important content in geographic information processing and plays an important role in the realtime processing of massive remote sensing images. Aiming at the problem of single node failure, low scalability and low efficiency in traditional remote sensing image processing, this paper proposes a distributed storage and query scheme of remote sensing data based on HBase. In this method, the remote sensing image is divided by uniform mesh, and an index scheme based on grid ID and Hilbert curve is designed according to the block information. Then, the HBase filter mechanism is used to design the filter column family so as to achieve the purpose of screening data in the query. In addition, the MapReduce parallel processing method is adopted to write and query image data. The experimental results show that, compared with MySQL and MapFile, the proposed method can improve the query speed of data and has good scalability.

   

  SDT algorithm and its improvement

  for historical data in SCADA

  XU Xudong，FU Yanping

  2018, 40(06): 999-1004. doi:

  Abstract ( 133 )   PDF (556KB) ( 205 )   　　 Review attachment

  With the rapid development of Internet of things and big data，the amount of data collected by the supervisory control and data acquisition (SCADA) system is growing exponentially, resulting in that the traditional swing door trending (SDT) algorithm can no longer meet the needs of the SCADA system for historical data compression. We propose an advanced swing door trending (ASDT) algorithm and implement it in Java language based on the deep research on the data compression method and the SDT algorithm particularly. The ASDT algorithm which uses sine curve fitting data to achieve data compression has better compression results in comparison with the traditional SDT algorithm. Experimental results show that compared with the traditional SDT algorithm，the ASDT algorithm can improve the compression ratio without significant increase in compression error.

   

   

  A robust zero-watermarking algorithm

  based on the block FRIT-SVD

   

  QU Changbo，YU Zhilong，LI Dongdong

  2018, 40(06): 1005-1016. doi:

  Abstract ( 120 )   PDF (1458KB) ( 237 )   　　

  Wavelet transform is not the best way for extracting watermarking image contour features. Compared with the wavelet transform, the ridgelet transform has better accuracy and sparse approximation performance. We propose a robust zerowatermarking algorithm based on ridgelet transform domain, which combines the twodimensional chaotic system, SVD and bitplane to construct the watermarking. Firstly, the algorithm takes advantage of the wavelet transform to extract the lowfrequency domain in the carrier image. Secondly, we decompose the lowfrequency domain according to the blocking strategy. When all the blocks are transformed by the FRIT, the characteristic matrix can be constructed by the SVD. Moreover, we apply the twodimensional chaotic system to encrypt the characteristic matrix. Finally, we extract the most significant bitplane in the feature matrix, and generate registration zerowatermarking information by combining the encrypted meaningful watermarking, which not only improves the robustness of the watermarking,
  but also greatly enhances the security of watermarking through double encryption. Experimental results demonstrate that the algorithm has good robustness, security and easy operation, and that it can effectively resist various types of attacks.

   

  An SDN-based load balancing algorithm

  in data center network

  FAN Zifu,ZHANG Dan,LI Shu

  2018, 40(06): 1017-1022. doi:

  Abstract ( 116 )   PDF (670KB) ( 183 )   　　

  Because the distributed architecture for traditional networks makes the load balancing technology difficult to meet the requirements of low cost, high flexibility and adaptive adjustment, a load balancing algorithm based on Software Defined Networking (SDN)in data center network is proposed to solve such problem. Firstly, according to the current load status of a path and its link load fluctuation,a weight is set for the path and as the basis for path selection.Secondly, a load balance degree is used to measure the network load status. Finally, for the flow that needs to be scheduled, its flow size range is further limited to ensure efficient flow scheduling. The simulation results show that, compared with other algorithms, the proposed algorithm can effectively improve the utilization ofnetworkresourcesandbalance the loadof the whole network.

   

  A certificateless proxy re-signature

  scheme with aggregate property

  YANG Xiaodong,YANG Ping,GAO Guojuan,LIU Tingting,WANG Caifen

  2018, 40(06): 1023-1028. doi:

  Abstract ( 126 )   PDF (431KB) ( 150 )   　　

  Most existing proxy resignature schemes are based on certificates or identity cryptosystems, and there are issues such as certificate management, key escrow security. In order to overcome the shortcomings such as strong security assumption and high computation cost in the existing proxy resignature schemes, a certificateless proxy re-signature scheme with aggregation property is proposed by combining proxy re-signature and certificateless public key cryptosystem. This scheme can aggregate an arbitrary-sized set of signatures or re-signatures into a set of signatures, and effectively reduce the communication overhead and computation cost of signature verification. The analysis results show that the proposal has a shorter signature length and resignature length. The proposed scheme is existentially unforgeable under the k-MCDH assumption.

   

   

  Local differential privacy protection and its applications

  GAO Zhiqiang，CUI Xiaolong,ZHOU Sha,YUAN Chen

  2018, 40(06): 1029-1036. doi:

  Abstract ( 257 )   PDF (659KB) ( 363 )   　　

  Privacy protection has become the focus of information security research. Differential privacy has been highly recommended by the theoretical community since 2006. In recent years, the local differential privacy in the industry crowdsourcing model has attracted much attention. We analyze the local differential privacy model from the perspective of theoretical research and engineering practice, and summarize the applications of local differential privacy theory in data collection and data analysis. In the aspect of data collection, the main research and application results of local differential privacy are introduced, and the methods are analyzed and compared from the perspective of differential privacy. In the aspect of data analysis, the implementation and analysis of local differential privacy in encoding, decoding and statistics are discussed, and these algorithms are analyzed theoretically. Finally, on the basis of indepth comparison and analysis of existing technologies, the challenges and research directions of local differential privacy techniques are summarized.

   

  Performance analysis for laregescale heterogeneous cellular

  networks with massive MIMO relays based on stochastic geometry

  ZHOU Meng1,JIA Xiangdong1,2,XIE Mangang1

  2018, 40(06): 1037-1044. doi:

  Abstract ( 114 )   PDF (1928KB) ( 151 )   　　

  By using stochastic geometry theorem, we investigate the spectral efficiency (SE) of largescale heterogeneous cellular networks (HetNets) with massive MIMO relays, where the maximum ratio combining/ maximal ratio transmission, zeroforcing processing scheme as well as amplifyandforward are employed at base stations and the macro cells are overlaid with dense small cells. To obtain the aggregate SE in target cell, we first formulate the signaltointerference noise ratios (SINR) for a user pair as well as the expression of power amplification factor. We then achieve the closedform expressions of aggregate SE. We find that when the radius of small cells is less than the guard radius we can acquire optimal (or suboptimal) system performance. For the guard radius, there is an upper bound, only under which the increase of the guard radius can contribute to the improvement of SE. Meanwhile, the radius of small cells has an optimal lower bound and only when the radius of macro cells is less than the optimal lower bound, the aggregate SE increases with the macro cell radius.

   

  PUF based authentication protocol in 

  mobile and largescale RFID systems

  LI Song1,SUN Ziwen1,2

  2018, 40(06): 1046-1053. doi:

  Abstract ( 121 )   PDF (779KB) ( 200 )   　　

   

  Aiming at the security problem in mobile radio frequency identification (RFID) systems, we employ the physical unclonable function (PUF) to study authentication protocol in mobile and largescale RFID systems. To solve the problem of impersonation attack to the reader, the Vaudenay’s model is extended by introducing the corrupting ability of the adversary to the reader and the reader is further authenticated by the server. To solve the problem of insufficient computation ability of tag and longtime cost of the server for  searching target tag, the PUF is adopted to generate the session key to reduce the computation cost of tag encryption, and the server quickly identifies the tag and the reader using the shared key XOR operations. Our analysis proves that the proposed security protocol can realize destructive privacy according to the Vaudenay's model theory. Simulation results show that the search time of the server in PMLS protocol does not increase with the number of the tag by using the proposed security protocol, which meets the application requirements of mobile and large scale RFID systems.

   

  An active defense technique based

  on network security awareness

  LIU Shiwen1,5,MA Duoyao2,4,LEI Cheng1,3,5,YIN Shaodong2,4,ZHANG Hongqi1,5

  2018, 40(06): 1057-1061. doi:

  Abstract ( 217 )   PDF (1008KB) ( 268 )   　　

  As a key technique to break through the bottleneck of passive defense, network active defense becomes a hotspot in network information security. To solve the blindness problem of hopping mechanism in the course of network defense, we propose a novel active defense mechanism based on network security situation awareness. Firstly, a network security situational awareness method based on scanning flow entropy is designed, which enhances the targeted defense by discriminating the adversary scanning strategy. Based on this, an active defense mechanism based on endpoint information transformation is proposed. It can increase the difficulty and the cost of attacks by randomly changing network topology dynamically through transforming endpoint information. Theoretical and experimental analyses show that the proposed active defense technique can be employed efficiently under different scanning strategies.

   

  A reference channel model of indoor integrated

  power line communication and VLC systems

   

  GAO Shuli1,2，ZHANG Jian1，YANG Hui2

  2018, 40(06): 1062-1066. doi:

  Abstract ( 136 )   PDF (532KB) ( 237 )   　　

  The design of the indoor integrated power line communication(PLC）and visible light communication(VLC) requires study on the reference model of integrated systems. Since the traditional analytical model of the integrated channel of PLC and VLC does not determine the parameters of the channel model, it can only describe general characteristics of the channel qualitatively. We propose a reference model of the integrated PLC and VLC systems through the actual measurement of the integrated channel in the typical scenario, and the reference model is determined by a small set of parameters. The model parameters are fitted based on the least squares method, and the practicability of the integrated channel is improved by the obtained parameters. We then set up a test network to verify the correctness of the model, and the results show that the proposed model can significantly describe transfer characteristics of the hybrid system.

   

  MCS calculation based on unsatisfiable reasons

  HU Xiaosa，ZHANG Yueling，LI Jianwen，PU Geguang，ZHANG Min

  2018, 40(06): 1067-1074. doi:

  Abstract ( 111 )   PDF (586KB) ( 201 )   　　

  The minimal correction subset (MCS) of a Boolean formula is a set of clauses. By removing the MCS from a given unsatisfiable formula, the new generated formula becomes satisfiable. For any clause in the MCS, keeping the clause in the given unsatisfiable formula results in a new unsatisfiable formula. Minimal unsatisfiable core, MaxSAT and maximal (minimal) partial assignment can be solved by solving the MCS and adjusting the set of constraints. The MCS can also be applied to practical problems such as fault localization, model checking and configuration optimization. We propose an MCS calculation algorithm based on unsatisfiable reasons returned from SAT solvers, and implement a corresponding tool named CUC. Comparing with the state-of-theart tool LBX, the CUC outperforms LBX. The CUC can solve 5% (65) more formulas on average, and it is 2.5 times faster than LBX.

   

  Algebraic specifications of service composition

  CHEN Ying1，LIU Dongmei1,ZHU Hong2,LAN Bin1,HE Juanjuan1

  2018, 40(06): 1075-1083. doi:

  Abstract ( 122 )   PDF (499KB) ( 132 )   　　

  Existing methods for the specification and description of service compositions cannot ensure the correctness of the compositions to be verified and tested. Addressing this problem, we propose an algebraic specification method. A package facility is introduced into the serviceoriented formal specification language SOFIA to support the method. In this approach, various types of entities in a serviceoriented system are specified by a collection of algebraic specification units, and each specifies one type of entities in the system. When multiple services are composited together, based on the algebraic protocol package of these services, on the one hand, the interaction process and semantics of the composite service are abstractly defined so as to form the implementation specification that describes the implementation mode of the service composition. And on the other hand, the external interface and functional semantics of the composite service are abstractly defined so as to form the abstract specification that describes the requirements of the service composition. Based on the dual structure of the implementation specification and the abstract specification, the “implementation” relationship that must be satisfied between the implementation specification and the abstract specification is further defined, and it is proved that satisfying the “implementation” relationship can ensure the correctness of the implementation, thus laying the theoretical foundation of verification and testability of service composition. Finally, a case study demonstrates the abstractness, expressibility and verifiability of using algebraic specification to describe service composition.

   

  ASM reduction for test generation

  YANG Yang,HE Liuliu,SHANG Ying,LI Zheng

  2018, 40(06): 1084-1092. doi:

  Abstract ( 89 )   PDF (858KB) ( 149 )   　　

  Model based web application testing is a crucial approach in software testing. The ASM model is based on the presentation layer of web applications, and presents interactive, dynamic and low coupling features of web applications via source code analysis. The test case generation based on the ASM model, allowing for testing the unexpected behavior of users, is expanded by adding the states of invalid access and the edges of invalid transition to the primary path. However, with the expansion of web applications, the space of test cases is exploded due to the increase of the states of invalid access and the edges of invalid transition. To solve the problem, the ASM model is reduced by merging the equivalent states and equivalent transitions, which are defined for ASMbased test generation through the research on the ASM. In this way, the states of invalid access and the edges of invalid transition are consequently reduced, and thus the test case space is effectively decreased. Experiments on a realworld web application show that the test case space of the ASM model is reduced by 74.38%  without decreasing the coverage of atomic sections and the number of detected errors.

  
   

  CPN model based  standard feature

  verification for REST service architecture

  ZHAO Yuqiang,LIU Jing

  2018, 40(06): 1093-1102. doi:

  Abstract ( 108 )   PDF (1293KB) ( 128 )   　　

  Nowadays, the representational state transfer (REST) service architecture is widely used in largescale and scalable distributed web systems. If the REST service architecture does not comply with its standard features, it can result in degraded performance or low scalability of the RESTbased web systems. Therefore, in order to enhance the quality of system designing, before implementing the web system based on the REST service architecture, it is necessary to verify whether the system design meets the standard features of the REST service architecture. In this paper, we propose a standard feature verification method for REST service architecture based on colored Petri nets (CPN) model. Firstly, five standard feature constraints of the REST service architecture are modeled using the CPN. Then a verification method is proposed based on synchronized matching of the execution paths in model state space, that is, considering both the CPN model of the application system and the CPN model of standard feature constraints, synchronized matching is performed on the respective execution paths in the model state space. If these paths can be executed synchronously, the application system meets the requirement of the REST standard feature constraints. We validate the usability and validity of the proposed verification method using a practical course management web system based on the REST service architecture. Experimental results show that the proposed method can effectively confirm whether the web application system design based on the REST service architecture conforms to the standard feature constraints of the REST service architecture. Besides, it can also provide intuitive and feasible execution data when the standard feature constraints are not met, which can facilitate the defects location and correction of the following design of application systems.

   

  A recognition-oriented unsupervised

  feature learning algorithm

   

  XIA Haijiao 1,2，TAN Yihua 1,2

  2018, 40(06): 1103-1110. doi:

  Abstract ( 95 )   PDF (909KB) ( 167 )   　　

  Feature extraction is a key part of  image recognition, and precise feature expression can generate more accurate classification. We improve the recognition rate of the singlestage computational structure by adopting soft threshold encoder and the orthogonalizing visual dictionary of the orthogonal matching pursuit (OMP) algorithm. Besides, we build a twostage computational structure which extracts images’ features and increases the recognition rate. Experiments demonstrate that adopting softthreshold encoder and the OMP algorithm can increase the ability of extracting features of the singlestage computational structure and enhance imagerecognition rate in bigsample datasets. The twostage computational structure can improve recognition rate on selfselection datasets. The OMP algorithm can improve recognition rate of the VOC2012 dataset. For selfselection datasets, the twostage computational structure outperforms the singlestage computational structure and network in network (NIN), and is equivalent to convolutional neural networks (CNN), indicating that the twostage computational structure is adaptive to selfselection datasets.

   

  An improved One-Cut interactive

  image segmentation algorithm

  WANG Dong，TANG Jinglei

  2018, 40(06): 1111-1118. doi:

  Abstract ( 163 )   PDF (1052KB) ( 216 )   　　

  As a typical interactive color image segmentation method, the Grabcut algorithm is an important technique in the field of computer image. However, with the arrival of big data age, the types and quantities of image data are increasing exponentially. The workload of image segmentation is significantly increasing, and a higher demand is raised for the efficiency of image segmentation algorithms. Aiming at the problem of low efficiency and accuracy of the GrabCut algorithm, we propose an improved OneCut interactive image segmentation algorithm. Firstly, the energy function is built with the OneCut L1  distance term to avoid the Nphard problem faced by the GrabCut algorithm. Secondly, the apparent overlap penalty in the energy function is improved and the network structure is optimized by the color histogram acceleration technique. Finally, the complexity of the network diagram is reduced and the image segmentation efficiency and accuracy are improved. Experimental results show that the improved OneCut interactive image segmentation algorithm can significantly improve the segmentation efficiency and segmentation accuracy and a better segmentation result is obtained.