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

    • 论文
      SSDKV:a fast key-value system based on SSD devices  
      MEI Fei,CAO Qiang
      2016, 38(07): 1299-1308. doi:
      Abstract ( 246 )   PDF (648KB) ( 356 )     

      KeyValue (KV) store systems use SSD device to improve the response speed of I/O. However, the existing KV store systems adopt the standard file system to handle storing data in SSD devices, so the physical characteristics of SSDs are invisible to applications and the I/O behaviors of applications are invisible to storage media, limiting the further performance improvement of KV store systems. Therefore, we  design a storage mechanism, which concerns both KV data patterns and SSD device characteristics.. By combining with LevelDB, a LSMTreebased KV store, we implement a SSD friendly KV system—SSDKV, which is a lightweight KV store and runs in user space. Besides the basic function of providing KV interfaces, SSDKV introduces an innovative ondisk data deployment. By reducing the extra procedures required by the traditional file systems to handle  KV data, and managing the SSD storage space in a more proper way according to KV data, SSDKV can improve the performance significantly on the KV store. Compared with other file systems, SSDKV can improve write performance by 4 times and read performance by 1.5 times.

      A SoC verification platform based on reusable stimulus generation mechanism 
      SU Yiduan,YU Zhiguo,GU Xiaofeng
      2016, 38(07): 1309-1315. doi:
      Abstract ( 110 )   PDF (903KB) ( 286 )     

      Traditional stimulus generation mechanism is laborintensive and hard to reuse in the design process of systemonachip (SoC), thus the simulation efficiency is affected seriously. To solve this problem, we construct a virtual SoC verification platform based on the reusable stimulus generation mechanism. The verification platform calls a port stimulus file through a reusable stimulus generation module, and the stimulus file is converted into the timing signals that correspond to the ports of the verification circuit. The function verification of universal synchronous/asynchronous receiver/transmitter, interrupt and timer demonstrates the observability, controllability and feasibility of the stimulus generation mechanism. The verification result analysis shows that it only needs to modify the firmware and port stimulus files when different function points are to be verified. As a result, the amount of code changes during the reuse of the verification platform is reduced, the flexibility and efficiency are improved, and the SoC verification time is shortened.

      A fast retrieval method for large scale images
      based on multiple Hash algorithms  
      TANG Xiaoman1,WANG Yunfei1,ZOU Fuhao1,ZHOU Ke2
      2016, 38(07): 1316-1321. doi:
      Abstract ( 133 )   PDF (594KB) ( 398 )     

      Hash technique is regarded as the most promising method of similarity search, which can be used in large scale multimedia data search. In order to solve the problem of low efficiency of data retrieval over large scale images, we propose a reverse index tree structure based on segmental Hash codes, and elaborate its implementation principles. In this structure, Hash codes are segmented. We design an inverted tree index structure for each section of Hash codes, and build the Hash index structure with the combination of the bloom filters. To further improve the accuracy of the retrieval results, we design a fusing algorithm which constructs the weighted undirected graph separately for the ranking results of multiple Hash algorithms. The fusion algorithm of the sorting list based on multiple Hash algorithms is described in detail by PageRank. Experimental results show that the structure of the inverted index tree based on segmental Hash codes can greatly improve data retrieval speed. Compared with the conventional single Hash algorithm sorting technique, the rank fusion technique for ordered lists of multiple Hash algorithms has obvious superiority in retrieval accuracy.

      A domain ontology  storage method and
      its applications based on HBase 
      WANG Hong,SUN Kang
      2016, 38(07): 1322-1329. doi:
      Abstract ( 132 )   PDF (721KB) ( 309 )     

      Based on the analysis of the civil aviation emergency management domain ontology and its storage features, we propose a domain ontology storage method based on HBase. The storage data of domain ontology metadata and the instance data are stored separately. We describe domain ontology classification, the metadata of attribute information, and the storage model of RDF instance data, as well as the current loading process of RDF data  based on MapReduce. Combining with practical applications, a basic graph pattern query algorithm of domain ontology based on HBase API is implemented. We analyze the experimental results in the Hadoop environment, which show that the proposed method can provide theoretical support and a new method for storing the huge data of civil aviation emergency management domain ontology.

      130X.2016.07.005An ensemble feature selection algorithm
      for high dimensional microarray data 
      SUN Gang1,2,ZHANG Jing1,3
      2016, 38(07): 1330-1337. doi:
      Abstract ( 165 )   PDF (514KB) ( 310 )     

      Feature selection algorithms are an important tool for microarray data analysis, thus their classification ability and stability are essential for data analysis. We propose an ensemble feature selection algorithm for high dimensional microarray data to compensate for the lack of information on a single gene subset. We firstly adopt the signal noise ratio method to select discriminative genes, and then generate relevant gene subsets by evaluating the correlation between the candidate gene and discriminative gene through conditional correlation coefficients. We finally integrate resemblant gene subsets through the ensemble learning technology. Experimental results show that in most cases the classification ability and stability of the proposed algorithm is superior to those that select only a single gene subset.

      A big data clustering algorithm based on local key nodes  
      CAO Yang,QIAN Xiaodong
      2016, 38(07): 1338-1343. doi:
      Abstract ( 106 )   PDF (458KB) ( 342 )     

      In order to find a reasonable network structure in big data, we present a local search algorithm suitable for big data. Aiming at the uncertainty of the initial nodes and the timeconsuming fitness function computation, we introduce key local nodes and improve the fitness function to reduce the time consumption. Experimental results show that compared with classical algorithms the time consumption of the improved algorithm does not change much in smallscale data networks but is less in largescale data networks, which demonstrates that the proposed algorithm is feasible and effective and can be applied to the clustering of large-scale data.

      Multicore NPU based TCP large receive offload    
      LI Jie,CHEN Shuhui
      2016, 38(07): 1344-1349. doi:
      Abstract ( 207 )   PDF (576KB) ( 322 )     

      The current development of the Ethernet technology is much faster than that of memory and CPU technologies, and the memory access and CPU processing network stack have become the bottleneck of TCP performance on end systems. The constantly increasing network bandwidth burdens  the CPU severely, and approximately 1GHz CPU resource is needed to process 1Gbps network traffic. We therefore take a multicore NPU as the NIC and the TCP's checksum verification and packets reordering functions are offloaded. Small TCP packets aggregated into fewer but larger packets by the multicore NPU, thus reducing both the number of packets processed by network stack and the number of interrupts generated by the NIC, and eventually improving the TCP performance on end systems. Experimental results show that 4.9 Gbps TCP receive data throughput can be achieved in a 10Gbps network.

      A serveraided verification signature
      scheme against collusion attacks         
      YANG Xiaodong,GAO Guojuan,LI Yanan,LU Xiaoyong,WANG Caifen
      2016, 38(07): 1350-1355. doi:
      Abstract ( 122 )   PDF (407KB) ( 295 )     

      Serveraided verification signature can effectively reduce computation cost of the verification of a digital signature, which is very suitable for lowpower devices. However, most serveraided verification signature schemes in the standard model are vulnerable to collusion attacks of the server and malicious signers. To improve the security performance of the serveraided verification signature scheme, we present a novel serveraided verification signature scheme, which is proved to be secure under both collusion attacks and adaptive chosen message attacks in the standard model. Analysis results show that the proposed scheme can effectively reduce the computation cost of bilinear pairing operation and the computation complexity of the signature verification algorithm. The proposed scheme is more efficient than the existing signature schemes in the standard model.

      A scene labeling algorithm for multiscale deep
      networks based on deep learning 
      MA Chenghu,DONG Hongwei
      2016, 38(07): 1356-1361. doi:
      Abstract ( 203 )   PDF (656KB) ( 497 )     

      There are two primary problems  in scene labeling: how to produce good internal representations of the visual information and how to use contextual information efficiently. To solve the two problems, we present a scene labeling algorithm for multiscale deep networks based on deep learning, a supervising model. Unlike traditional multiscale methods, the model contains two deep convolutional networks: one takes the global information into account and extracts the lowlevel features of the largescale image; and the other one combines the local information of the image with the low level features, and obtains a set of dense and complete image features, thus a powerful representation that captures texture, color and contextual information is achieved. Compared with many standard approaches, the proposal does not depend on segmentation technique or any task specific features. Our approach yields good performance on the Stanford Background Dataset.

      TBMC:a tracking strategy based on
      multidetector collaborating in DTNs  
      DENG Yiqin,ZHAO Ming,TANG Fengxiao,CHEN Zhigang
      2016, 38(07): 1362-1368. doi:
      Abstract ( 107 )   PDF (550KB) ( 295 )     

      Target tracking problem in DTN networks is a widespread concern, however, the current strategies mainly focus on a single detector tracking a single destination. Some strategies use the possibility that nodes encounter, some take advantage of the location information of nodes, and some adopt more innovative tools, such as the mobile location information and the time left by the target node. These policies make a direct or indirect use of the information of the destination node, however, the general not very high success rate makes their applications a subject to certain restrictions. We therefore present a distributed target tracking strategy based on multidetector collaborating named TMBC, which achieves multidetector collaborative tracking by the remote detection mechanisms DTRs and the close cooperative data exchange mechanism TRs. It can avoid conflict and redundancy detection to improve tracking efficiency. Theoretical analysis and experimental results verify that the proposal can effectively reduce the tracking steps to improve the success rate in comparison with the traditional methods.

      An energybalanced routing algorithm based on games model       
      CAI Zhao1,MA Linhua1,HUANG Shaocheng1,ZHANG Song1,TIAN Yu2
      2016, 38(07): 1369-1375. doi:
      Abstract ( 133 )   PDF (621KB) ( 275 )     

      Aiming at the unbalanced energy of the sensor network and the vulnerability of the network performance to selfish nodes, we establish a balanced game model of energy consumption (EBGM) algorithm based on the game theory, which encourages nodes' cooperative behavior. We introduce energy concern factors, which adjusts forwarding willingness according to the proportion of the existing energy differences between the node and its neighbors, rather than the mode of the traditional game algorithms which regard residual energy as the only standard to adjust forwarding willingness. We analyze the theory of the EBGM algorithm, and prove the existence of the Nash equilibrium that tends to be Pareto optimal equilibrium. Simulations indicate that the EBGM algorithm can promote cooperative behavior of nodes while balancing energy cost and prolonging the lifetime of the whole network.

      A RFID mutualauthentication protocol with synchronous
      updatedkeys based on chaotic encryption       
      HU Yingmeng,ZHANG Xiaohong
      2016, 38(07): 1376-1383. doi:
      Abstract ( 135 )   PDF (659KB) ( 285 )     

      Since the security privacy issues of RFID systems are increasingly prominent, we propose a RFID mutualauthentication protocol based on chaotic sequence encryption to tackle those problems. Chaotic sequence can be generated to encrypt the keys by making full use of the sensitivity of chaos to the initial value. An update mechanism of the dynamic tag keys is introduced and a selfsynchronized scheme is designed, which achieves a second verification for the tags. The security of the protocol is verified by the BAN logic, and it is compared with the existing schemes in terms of security and performance. The results show that the proposed protocol can greatly decrease the cost of tags, reduce the amount of computation of tags and the backend database, and enhance the search efficiency of the backend database. It can not only effectively solve the privacy and security issues of the RFID system, but also improve the efficiency of the RFID system, thus being more suitable for the low cost RFID system.

      Combination of coding overhead and network security in the network coding optimization scheme   
      XU Guangxian,YANG Dongli,GAO Song,XU Chunyan,JIN Yubo
      2016, 38(07): 1384-1390. doi:
      Abstract ( 112 )   PDF (645KB) ( 270 )     

      Network coding technology has great advantages in improving network throughput and transmission efficiency. However, this technique requires additional coding operation and increases coding overhead. In order to reduce coding overhead while taking into account the network coding security level under the guarantee for maximum multicast rate premise, we present a program that combines coding overhead and network security in the network coding optimization. Firstly, this program adds a preprocessing mechanism to the niche genetic algorithm that bases on a preselection mechanism. We then construct a new fitness function, and a more scientific way is adopted to determine the size of population. Simulation results show that the proposed algorithm outperforms the algorithms that base on the traditional genetic algorithm in terms of convergence time, evolution of algebra, coding overhead and security.

      Traffic congestion control strategy under different  information sharing degrees         
      LI Yong1,2,CAI Meng si1,LI Li1,ZOU Kai1
      2016, 38(07): 1391. doi:
      Abstract ( 104 )   PDF (659KB) ( 127 )     

      Information sharing degree is an important factor that affects traffic efficiency, we therefore analyze traffic congestion phenomenon and its characteristics in three information sharing patterns, which are information blocking pattern, local information sharing pattern and global information sharing pattern. We then exploit the dynamic network node behaviors, and build a corresponding traffic congestion propagation model through the method of probability generating function, branching process and coordination game. We finally obtain the critical value of traffic congestion propagation, and make a comparison of traffic congestion control strategy in different information sharing patterns. Simulation results show that when traffic network flow is relatively small, the information blocking pattern and the local information sharing patterns are more effective, while the global information sharing pattern is conducive to  controlling largescale congestion though there are more controllable difficulties.

      Application of multiobjective genetic algorithms  based on clustering in class responsibility assignment        
      LI Ya jin1,LIU Wei1,2,HU Zhi gang1
      2016, 38(07): 1398-1404. doi:
      Abstract ( 106 )   PDF (525KB) ( 310 )     

      In the process of objectoriented software design and implementation, responsibility assignment problem (CRA) is one of the most important and complicated procedures, which affects the quality of software to a large extent. In order to achieve the goal of CRA automatically, we propose a CRA multiobjective optimization model which is built from the perspective of cohesion and coupling metrics. On the basis of fast nondominated sorting genetic algorithm, we introduce the agglomerate hierarchical clustering technology to ensure population diversity and to avoid premature convergence. Experiments on automatical class responsibility assignment verify the correctness of the algorithm, whose results are also compared with an existing welldesigned software system. In addition, compared with the single objective genetic algorithm and the SPEA2 algorithm, the proposed algorithm has the best CRA operation effect.

      Weighted regular grammars over valuation monoid  
      ZHAO Fei,LI Yong-ming
      2016, 38(07): 1405-1412. doi:
      Abstract ( 123 )   PDF (398KB) ( 248 )     

      Regular grammars are necessary tools for the analysis of automata.  We introduce the concepts of weighted regular grammars and weighted similar regular grammars over valuation monoid, discuss the relationships among weighted regular grammar, weighted similar regular grammar and weighted finite automata (WFA). We show that for a given weighted regular grammar or a weighted similar regular grammar, their there is a WFA which is equivalent to the weighted regular grammar or the weighted similar regular grammar.  We define the concept of distributable valuation monoids  and prove that on a distributable valuation monoid  for a given WFA  there is a weighted regular grammar or a weighted similar regular grammar, and they generate the same languages. Examples show that the distributivity of valuation monoid is not a given WFA  and we can find a weighted regular grammar or a weighted similar regular grammar which is equivalent to the WFA.

      Construction of domain ontology knowledge base for assembly tolerance synthesis   
      ZHONG Yan-ru,LU Hong-cheng,ZENG Cong-wen
      2016, 38(07): 1413-1418. doi:
      Abstract ( 99 )   PDF (802KB) ( 362 )     

      To reduce the uncertainty in mechanical product design and to solve the problem of effective sharing and smooth transferring of assembly tolerance information among heterogeneous CAX systems, we construct an ontology knowledge base for the assembly tolerance synthesis. We first analyze the assembly tolerance knowledge based on the ontology that contains abundant semantic knowledge and semantic structures. The web ontology language (OWL) is used to define the concepts and relations, and the semantic web rule language (SWR) is used to define the constraint conditions and distribution experience. Afterwards, the OWL-based structure knowledge is mapped to facts, the SWRL-based constraint knowledge is mapped to rules, and an ontology knowledge base for the assembly tolerance synthesis is constructed based on the inference engine. Meanwhile a prototype system of the assembly tolerance synthesis is developed based on the ontology, which can automatically generate the assembly tolerance types and assembly tolerance values.

      Research and implementation of the second class  activity management platform based on workflow       
      ZHAO Dong,ZHANG Wen-ning,CHE Zhan-bin
      2016, 38(07): 1419-1424. doi:
      Abstract ( 107 )   PDF (861KB) ( 307 )     

      The second classroom activities have a unique advantage in cultivating innovative cap abilities in colleges and universities. It is necessary to use modern information technology to promote the dynamic processes and optimize the rate of resource sharing. We thus propose an enhanced role-based control (RBAC) model and the workflow of competition and public benefit activities are modeled based on the Petri net. To improve the reliability and safety of the system, the workflow engine is developed using the .Net technology. Furthermore, the second classroom activities management platform based on workflow in colleges and universities is analyzed and designed.

      A cross-class query optimization method of object deputy database        
      JIANG Lian,LI Rong-rong,PENG Zhi-yong
      2016, 38(07): 1425-1433. doi:
      Abstract ( 146 )   PDF (726KB) ( 336 )     

      Cross-class query in object deputy database can make full use of object deputy model’s flexibility to provide users with personalized data services, and its efficiency is very important. However, using existing path-expression based methods to handle cross-class query with multiple attributes, the query performance degrades due to redundancy of common path nodes. To solve the problem of terminal object's quick acquisition in cross-class query, we propose an optimization approach to reduce the repeated and unnecessary traverse of common node objects in the path. Our approach consists of two key strategies. The first strategy is to consider the whole path nodes as a virtual path view, so we can obtain node objects uniformly and avoid redundant traverse of common path nodes in multiple cross-class attributes query. The second strategy aims to materialize common path nodes to handle cross-class query with a too long path. We select the origin object and the terminal object of materialized query according to the cost estimation strategy. Particularly we leverage cache to reduce the traverse of intermediate nodes. Comparative experiments on real data regarding both the number of attributes and the scale of result sets verify the effectiveness of our optimization method.

      An environmental health warning system for pension communities 
      LI Zhi-shan1,2,SHI Run-he1,2
      2016, 38(07): 1434-1439. doi:
      Abstract ( 101 )   PDF (756KB) ( 381 )     

      Current environmental issues draw more attention and cause a great threat to older age groups who have weak physique. Combining modern information technology with pension communities can provide dynamic and personalized environmental health information services for the old. In addition, it makes pension communities much smarter. We therefore design a warning system based on Java Web to dynamically forecast disease indexes including cold, high blood pressure, chronic bronchitis and some other diseases. Through the demand analysis of and the functional design for different roles, including the old, family members, medical staff and administrators, this system can realize medical staff management, remind the old timely and send notification to family members. At the same time, it is not only applicable to pension communities, but also to the centralized management of the old at home.