Author Login Editor-in-Chief Peer Review Editor Work Office Work

15 November 2019, Volume 45 Issue 11
    

  • Select all
    |
  • ZHANG Chaodong, WANG Baosheng, DENG Wenping
    Computer Engineering. 2019, 45(11): 1-8. https://doi.org/10.19678/j.issn.1000-3428.0055286
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The current supply chain traceability system based on centralized design has many disadvantages.Besides its performance improvement depending on the central server,other problems such as difficulty in supervision and accountability,excessive intermediate circulation nodes and opaque data also exist.Therefore,we design and implement a supply chain traceability system based on side chain technology.The system can achieve the goods management,information sharing and product traceability through Ethereum smart contract.At the same time,we use the side chain technology to expand the Ehtereum to meet the needs of practical applications.Experimental results show that the proposed system can trace the whole life circle of products in the supply chain,facilitating user verification query and third-party supervision.Besides,the throughput of the system reaches 200 TPS,and the cost of each transaction is reduced to 0.07 yuan.
  • LAO Weilun, WANG Baiyong, ZHANG Rui, WANG Jiabei
    Computer Engineering. 2019, 45(11): 9-15. https://doi.org/10.19678/j.issn.1000-3428.0053925
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Due to the large volume,various types and high security requirements of power data,its disaster tolerance and backup scheme should provide scalability and privacy protection.Blockchain is tamper-resistant,decentralized and traceable,which caters to such requirements.Therefore,this paper proposes a new blockchain-based disaster recovery and backup scheme of power data,integrating AONT,threshold secret sharing and other cryptographic technologies.The prototype system is implemented on an open-source blockchain platform,Ethereum.Experimental and Analysis results show that the scheme can ensure the consistency,tamper resistance and confidentiality of backup data.It is highly scalable,reducing the construction cost of infrastructure as well as the risk of single point of failure.
  • ZHANG Bin, GUANG Hui, CHEN Xi
    Computer Engineering. 2019, 45(11): 16-23,31. https://doi.org/10.19678/j.issn.1000-3428.0054940
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To improve the security of Wireless Mesh Network(WMN),this paper proposes a Smart Contract-based Security Architecture(SCSA) that integrates blockchain technology.The proposed architecture builds blockchains on routing nodes.Three types of smart contracts are deployed to provide hierarchical management for public keys of nodes,including public key updating,public key cancellation and user access authentication.Also,the Elliptic Curve Diffie-Hellman(ECDH) and symmetric encryption are used to implement secure communication between nodes.Experimental results show that SCSA can efficiently resist node clone attacks or node forgery attacks,key compromise attacks and denial of service attacks,ensuring the security and efficiency of network communication.
  • YE Jueyu, YUE Qiaoli, WANG Qian, ZHANG Haikuo
    Computer Engineering. 2019, 45(11): 24-31. https://doi.org/10.19678/j.issn.1000-3428.0055140
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Traditional defense system based on Domain Name System(DNS) is vulnerable to saturated traffic attacks and DNS hijacking attacks.To address the problem,this paper performs theoretical analysis and research on key factors in preventing network attacks,and constructs a model for network flows and state transitions.On this basis,a DNS collaborative defense system based on hyperledger is proposed.In the system,multi-lateral resources are integrated through consortium blockchains for collaborative resistance to network attacks.The channel architecture and endorsement policies of hyperledger are used to implement privacy protection and network information sharing.Analysis results show that the collaborative defense system is highly safe at the data level and the service level,providing a feasible solution to security threats in the domain name industry.
  • FANG Yi, DENG Jianqiu, CONG Linhu, LIU Chongyi
    Computer Engineering. 2019, 45(11): 32-36. https://doi.org/10.19678/j.issn.1000-3428.0055794
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The alliance chain is a kind of blockchain that allows authorized nodes to join in the network.However,when the network condition is poor,the nodes will join and exit dynamically.Therefore,on the basis of ring signature theory,ElGamal digital signature algorithm and PBFT algorithm,we propose a ring signature scheme based on ElGamal digital signature algorithm to cope with this problem.First,we analyze the correctness and anonymity of the ring signature algorithm.Then,we use the ring signature scheme to improve the signature and verification process of the PBFT algorithm.Finally,we adopt Caliper,the blockchain performance testing framework in Fabric,to test the performance of the improved scheme.Experimental results show that the improved,ring signature-based PBFT consensus algorithm can solve the dynamic joining and exiting problem of nodes in the network while reaching the fault-tolerant rate of the Byzantine node of the original PBFT algorithm,which has certain practical values.
  • WANG Haiqi, ZHAI Wenlong, YAN Bin, FEI Tao, LI Xuewei, CHEN Haibo, LI Jian
    Computer Engineering. 2019, 45(11): 37-46. https://doi.org/10.19678/j.issn.1000-3428.0053022
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    As the moving trend of an object is reflected by its driving distance and forward motion direction,this paper proposes a trajectory similarity measurement method based on motion direction.The features of motion direction are measured with actual average direction,linear average direction and longest common direction sequence.The method also considers the features of the shape and moving distance,which are measured with the area of regions bounded by trajectories.The optimal combination of the shape,moving distance and motion direction are taken as the final form of trajectory similarity measurement.The method is applied to GPS trajectory data clustering of occupied taxis in Beijing,and results show that the similarity measurement method can effectively distinguish moving trends and directions of objects with a higher clustering accuracy rate than the longest common subsequence method.
  • CHEN Junxiao, LI Zhongsheng, LIU Yimin, LI Qiuhong, WANG Wei
    Computer Engineering. 2019, 45(11): 47-53. https://doi.org/10.19678/j.issn.1000-3428.0054487
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the problem of poor performance of massive time series data query based on time series index of unbalanced tree,a DB-DSTree index based on MapReduce is proposed.The balanced time series index DHD is used as a routing tree to create a distributed DSTree,and the data locality in batch queries is fully mined.Similar queries are routed to local nodes to improve the non-balance property of DSTree index.Experimental results show that DB-DSTree index has good balance and locality,it can reduce the query range of subtree and I/O time of disks,and improve the query efficiency.
  • WANG Dafei, XIE Wujie, DONG Wenhan
    Computer Engineering. 2019, 45(11): 54-61. https://doi.org/10.19678/j.issn.1000-3428.0054988
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The Extreme Learning Machine(ELM) based on cost-sensitive learning has its advantages in dealing with imbalanced data classification problems.However,it fails to consider the distribution characteristics of samples in different classes and the importance of each sample in the same class,both of which can have influence on the classification results.Therefore,we propose a setting method for misclassified penalty factor based on the proportion of sample size.Besides,based on Mini-batch k-means clustering and distance measure,we propose a determination method for the weights of samples in the same class.On this basis,we build the output matrix of the hidden layer to distinguish the positive and negative categories.According to the relationship between the size of training samples and the number of nodes in the ELM hidden layer,we calculate the connection weights between the hidden layer and the output layer of ELM in two conditions,thus reducing the time complexity of the algorithm.Experimental results show that compared with ELM,WELM and other algorithms,the proposed algorithm has higher G-mean and F1 classification performance index.
  • ZHANG Shutao, TAN Haibo, CHEN Liangfeng, Lü Bo
    Computer Engineering. 2019, 45(11): 62-67. https://doi.org/10.19678/j.issn.1000-3428.0053439
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Traditional load balance methods for distributed crawlers fail in providing comprehensively efficient evaluation of crawler node loads,as they consider only a small number of affecting factors in load.Thus the tasks are not reasonably assigned.To address the problem,this paper proposes an efficient load balance strategy for distributed crawlers.The strategy analyzes affecting factors in the running time of crawler nodes,and uses BP neural network to construct a non-linear running time model based on multiple affecting factors for distributed crawler nodes.The model predicts the running time of each sub-node,and the minimum variance of the running time is taken as the target function of load balance strategies.The target function is resolved by using improved particle swarm optimization algorithm with constraints to form a task assignment scheme with balanced loads.Experimental results show that the load balance strategy can efficiently reduce the running time of distributed crawlers while meeting the high performance requirements of crawler nodes.
  • WANG Zehua, KE Xinsheng
    Computer Engineering. 2019, 45(11): 68-73,80. https://doi.org/10.19678/j.issn.1000-3428.0055033
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Most of the current recommendation systems have many defects,such as low recommendation accuracy,being subject to sparsity and poor stability,so we propose a recommendation algorithm based on Coclus joint clustering and non-negative matrix decomposition.Firstly,through Coclus joint clustering,we use the graph modularity maximization to divide row and column of the scoring matrix into g classes respectively,forming g×g low rank scoring submatrices through row and column transformation.Then we perform matrix decomposition on each low rank scoring submatrix and fill in the missing values to improve the recommendation quality.In the matrix decomposition stage,we adopt an improved non-negative matrix decomposition algorithm,introducing L1 and L2 norms respectively to improve the feature value selecting ability and prevent the over-fitting of model.Finally,we use the iterative algorithm of coordinate descent method to update the parameters.Experimental results show that compared with the baseline algorithm,the proposed algorithm has higher recommendation accuracy and better stability.
  • SUN Li, SU Yu, ZHANG Chi, ZHANG Tao
    Computer Engineering. 2019, 45(11): 74-80. https://doi.org/10.19678/j.issn.1000-3428.0052959
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    HRC code is an erasure code with high storage efficiency and low computation complexity,but it has some shortcomings,such as high computational overhead and complex implementation.So,we propose a new erasure coding type,the HRCSD code based on the optimized decoding algorithm of HRC code.The HRCSD code adopts an inner and outer layering structure.The inner redundancy is composed of the coding structure of the HRC code,while the outer layer adopts the offset copy strategy to rotate the original information and store it for parallel reading and writing.Experimental results show that compared with the three-copy technology and the S2-RAID erasure code,the HRCSD has higher fault-tolerant performance and lower repair overhead,which can satisfy the fault-tolerant requirements of large-scale distributed storage systems.
  • ZHAO Jihong, WANG Wenke, QU Hua, XU Xiguang, YAN Feiyu, YAN Haoliang
    Computer Engineering. 2019, 45(11): 81-85. https://doi.org/10.19678/j.issn.1000-3428.0052998
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The spectrum sensing scene with dynamic arrivals of Primary User(PU) in Cognitive Radio Network(CRN) often suffers from Laplace impulse noise,which leads to a drop in spectrum sensing performance.To address this problem,this paper proposes a spectrum sensing algorithm based on Absolute Value Cumulation(AVC).The algorithm assumes that received PU signals follow the Poisson distribution,and performs AVC on received signals to reduce interference of impulse noise.The accumulated sum of processed signals serves as the decisive statistics,and the mean value and variance of the decisive statistics are obtained.On this basis,the expression of decision threshold theory can be derived to judge whether the PU dynamically arrives,so as to implement spectrum sensing.Theoretical analysis and simulation results show that the proposed algorithm achieves a higher detection probability than the Improved Energy Detection(IED) algorithm under different false alarm probabilities,signal to noise ratio and sampling points.
  • HAO Weiqiang, XIONG Shuming
    Computer Engineering. 2019, 45(11): 86-90,96. https://doi.org/10.19678/j.issn.1000-3428.0052828
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In the routing protocols of Vehicular Ad Hoc Network(VANET) based on Road Side Unit(RSU),RSU updates data at a regular basis,causing a high risk of data invalidation in the Traffic Control Center(TCC).The resultant data packet loss can have a significant influence on the routing protocol performance.To address the problem,this paper proposes a fuzzy logic-based adaptive scheme to calculate the data update cycle of RSU in VANET.The scheme enables RSU to design a fuzzy logic-based control system,which considers the movement of vehicle nodes within the service scope of RSU and takes the data volume and data update speed in the registry as two inputs.The system can adaptively calculate the data update cycle to improve data validity in TCC.Experimental results show that compared with the BUS-VANET routing protocol,the VANET routing protocol based on this scheme can improve the stability and reliability of communication channels,and reduce the number of data update packets and transmission delay.
  • LI Shi, SHI Guyu, QIAN Yuwen
    Computer Engineering. 2019, 45(11): 91-96. https://doi.org/10.19678/j.issn.1000-3428.0052269
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Aiming at the problem of poor communication stability and high transmission energy consumption in the deployment of Power Line Communication(PLC) in smart power grid systems,a PLC network based on caching mechanism is proposed.The Multiple Input Multiple Output(MIMO) PLC communication mechanism is introduced,so that multiple network access nodes can simultaneously send data to one user,thereby improving the stability of network communication.An optimal precoding algorithm is designed to solve the problem of repeated data transmission on the backhaul link,thereby minimizing the energy consumption of users downloading files from cache nodes and servers.Simulation results show that the PLC network can effectively improve communication reliability and save system resources.
  • FAN Qiaoling, JIA Xiangdong, CHEN Yuwan, XU Wenjuan
    Computer Engineering. 2019, 45(11): 97-101,106. https://doi.org/10.19678/j.issn.1000-3428.0052851
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the imbalance problem of Uplink(UL) and Downlink(DL) in Heterogeneous Network(HetNet),we study a multilevel HetNet which combines Dual-Connectivity(DC),Decoupled UL and DL Association(DUDA),and propose a DUDA-DC Non-Orthogonal Multiple Access(NOMA) scheme.In this study, the total bandwidth is shared by the primary links and secondary links,and the receiver uses continuous interference elimination technology to separate the signals.First,with the help of the random geometry math toll,the object in the network is modeled as a Poisson Point Process(PPP).Then we study the coverage probability and traversal rate of Orthogonal Multiple Access(OMA) and Non-Orthogonal Multiple Access(NOMA) in the decoupled association multi-tier HetNet,and finally get the mathematical analytical expression.Simulation results show that compared with the OMA scheme,the NOMA scheme can effectively improve the network performance.
  • ZHAO Baishan, LIU Yongqiang
    Computer Engineering. 2019, 45(11): 102-106. https://doi.org/10.19678/j.issn.1000-3428.0053177
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the symbol synchronization problem of portable wireless video communication under multipath fading channels,this paper proposes an improved Orthogonal Frequency Division Multiplexing(OFDM) synchronization algorithm based on the lagged correlation and conjugate symmetry of symbol sequences.The algorithm uses the Schmidl & Cox algorithm to determine the synchronization scope of symbols,and uses the Park algorithm for forward search with the first point exceeding the threshold as the symbol synchronization point.Simulation results show that when the Signal Noise Ratio(SNR) is higher than 3 dB,the algorithm can implement accurate OFDM symbol synchronization under 6-route typical urban channels with a mean square error of frequency offset estimation for carrier wave smaller than 0.000 4.It is applicable to portable wireless video communication system.
  • WANG Yunming, CHEN Bo, ZHOU Min, ZHANG Yunhe
    Computer Engineering. 2019, 45(11): 107-111,120. https://doi.org/10.19678/j.issn.1000-3428.0052898
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the problem that the current heterogeneous wireless network access selection algorithm has difficulty in selecting the best wireless network from the alternative networks,we propose a new heterogeneous wireless network access algorithm.First,we use the fuzzy hierarchy analysis method to calculate the subjective weight of the attribute index of heterogeneous wireless networks.Then,we use the rough weight to calculate the objective weight.Finally,according to the service requirement,we dynamically adjust the subjective and objective weights to select the best access network.Experimental results show that the proposed algorithm can reduce the network switching frequency and congestion occurrence probability,effectively balance the network load and perform better in selecting the best access network.
  • GU Jing, HOU Yongping, ZHANG Yuxuan, ZHANG Xin
    Computer Engineering. 2019, 45(11): 112-120. https://doi.org/10.19678/j.issn.1000-3428.0054857
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In a two-layer heterogeneous network,when the pico base station performs load balancing on the network by using the cell range extension technology,the communication of the edge users of the pico cell is greatly interfered by the MBS base station.Therefore,this paper proposes an improved HSARSA(λ) algorithm based on heuristic functions.The RP-ABS sub-frame technology with reduced power reduces the interference of the MBS on the pico edge users while ensuring the communication performance of the macro base station,and uses the HSARSA (λ) algorithm improved by the heuristic function to interact with the environment.The sub-frame density and power level are configured for interference coordination.Simulation results show that compared with the original SARSA and Q-Learning algorithms,the pico edge user throughput is increased by 12% and 40% respectively,and the system user throughput is increased by 10.3% and 20.2% respectively.The communication performance of the edge users of the pico cell is significantly improved.
  • YIN Qingshan
    Computer Engineering. 2019, 45(11): 121-125. https://doi.org/10.19678/j.issn.1000-3428.0052762
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the high traffic and low precision of the Slice Mix Aggregate(SMART) data fusion algorithm applied to Wireless Body Area Network(WBAN),this paper proposes an improved SMART algorithm that optimizes data segmentation,collusion and fusion in SMART.During data segmentation,the algorithm prohibits non leaf nodes from transmitting data and constructs a node set that does not participate in data segmentation,so the traffic is reduced.During data collusion,data of negative segmentation is used to offset the loss in communication collision and improve the precision of data fusion. Finally,during data fusion,duplicate data in WBAN is eliminated for further traffic reduction.Performance analysis and simulation results show that the improved SMART algorithm can efficiently increase the computational efficiency and precision of data fusion while keeping the privacy protection performance of the SMART algorithm.
  • WU Ruixue, PENG Changgen, LIU Botao, DING Hongfa, XIE Mingming
    Computer Engineering. 2019, 45(11): 126-132. https://doi.org/10.19678/j.issn.1000-3428.0052660
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the problem that most privacy protection algorithms can not balance the data accuracy and data privacy protection degree,an anonymity algorithm based on association rules is proposed according to the association relation between quasi-identification attributes and sensitive attributes in data sets.The Aprior algorithm is used to establish the association rules between attributes,and the mutual information is used to measure the degree of association,which provides a basis for the classification of quasi-identification attributes.Also generalized boundaries and weights are set to avoid large anonymity costs.Experimental results show that the algorithm can reduce data loss and achieve the balancing between data utility and privacy protection.
  • YE Qing, YANG Xiaomeng, ZHAO Zongqu, ZHANG Jing
    Computer Engineering. 2019, 45(11): 133-137,143. https://doi.org/10.19678/j.issn.1000-3428.0052811
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Most existing forward secure proxy signature schemes are constructed based on traditional number theory problems,such as large integer decomposition,discrete logarithm and bilinear pairs,and they are easy to be cracked in a quantum computing environment.In order to deal with the threat of quantum computing environment and reduce the risk of secret key leakage in proxy signature,an identity-based forward secure proxy signature scheme with NTRU lattice is constructed.Introduce forward secure features into the proxy signature over the lattice.Under the random prediction model,the security of the program is converted to the intractability of SIS,and a detailed efficiency analysis is also given.Analysis results show that the scheme realizes forward secrecy and strong non-repudiation,and it is verifiable and strongly identifiable.The computing efficiency is also improved as a result of the shortened signature key of the proxy signer,and shortened signature and signature key of the original signer.
  • DENG Yuxin, TANG Zhangguo, ZHANG Jian, LI Huanzhou
    Computer Engineering. 2019, 45(11): 138-143. https://doi.org/10.19678/j.issn.1000-3428.0052521
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The carrier of traditional network covert channels is limited to the Internet protocol.To address this problem,based on the research of the Message Queuing Telemetry Transmission(MQTT) protocol at the application layer of Internet of Things(IoT),this paper proposes a convert channel model based on the MQTT protocol to migrate the carrier of network covert channels to the IoT platform.A block coding algorithm is designed by combining the quantity and sequence of commands,and the information to be sent is embedded in the MQTT session process.Experimental results show that the proposed covert channel has a high sneakiness and robustness.Compared with the covert channels constructed by the direct coding and sequence coding methods,the proposed convert channel has a higher transmission rate.
  • DUAN Xiaoyi, CHEN Dong, GAO Xianwei, FAN Xiaohong, ZHOU Yukun
    Computer Engineering. 2019, 45(11): 144-151,158. https://doi.org/10.19678/j.issn.1000-3428.0052538
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    According to the characteristics of the power curve of crypto chips,four kinds of machine learning algorithms including Support Vector Machine(SVM),Random Forest(RF),K-Nearest Neighbor(KNN) and Naive Bayes(NB) are analyzed and studied,and the optimal algorithm for power analysis attack is selected.For the data selection problem of the machine learning algorithm,the 10-fold cross-validation results of multiple sets of data sets with the same number but different constituent elements are used to select the model,which improves the test fairness and the generalization ability of the test results.In order to avoid the problem that the test set error is not enough to approximate the generalization error during the 10-fold cross-validation process,the four kinds of machine learning algorithms are evaluated by the combination of Friedman test and Nemenyi post-hoc test.The results show that the SVM is the optimal machine learning algorithm for power analysis attack.
  • LI Zhijia, XIA Wei
    Computer Engineering. 2019, 45(11): 152-158. https://doi.org/10.19678/j.issn.1000-3428.0053159
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The embedding capacity of the current difference histogram shifting algorithm is low.To cope with this problem,we propose a large-capacity reversible data hiding algorithm in encrypted domain.First,we use the single homomorphic encryption to enable a direct operation on the images in the encryption domain.Then,we divide the images into blocks of the same size and group the pixels of each image according to their positions.Finally,we establish a block difference histogram through the difference of four adjacent non-overlapping pixels,and choose the edge value of both sides as the embedding location to improve the embedding capacity.Experimental results show that the proposed algorithm can ensure that the directly decrypted image has a higher PSNR value when the information embedding rate is improved,and the carrier images can also be restored completely.
  • ZENG Fanyi, JING Xiaochuan, SUN Yunqian
    Computer Engineering. 2019, 45(11): 159-165. https://doi.org/10.19678/j.issn.1000-3428.0052658
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In order to solve the problem of information security exchange between industrial control systems and confidential working networks,and improve the efficiency of information exchange,a data exchange technology based on one-way transmission protocol is proposed.A data exchange trusted verification model is built to separate the business data into two one-way links,and the one-way isolation device is used to realize network security isolation and one-way information transmission.The information control management sub-platform is used to prevent the flow of high security information to low security domain,and the information collection management sub-platform is used to resist the attack on the classified information system.The data security transmission mechanism is designed,and data is packaged in private protocol.Analysis results show that the proposed technology can realize secure interconnection and data trusted exchange between industrial control system and classified information systems.
  • WANG Lian, ZHANG Zhao, ZHANG He, ZHANG Xunyang
    Computer Engineering. 2019, 45(11): 166-171. https://doi.org/10.19678/j.issn.1000-3428.0052845
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The wireless networks suffer from many security issues,such as pollution attacks,eavesdropping attacks and other diverse modes of attacks.To solve this problem,we introduce two parameters-generation identifier and time stamp into the wireless networks,and propose a RSA-based secure network coding scheme against multiple attacks.The scheme uses homomorphism to generate valid signatures for combined messages,so as to reduce the bandwidth consumption of the system.We compare this scheme with the network coding homomorphic signature scheme against inter-generation pollution attacks,and the results show that the cost ratio of the two schemes is close to 1,but the proposed scheme can resist intra-generation pollution attacks,inter-generation attacks and replay attacks at the same time.
  • WANG Xiaoying, XIE Jun, TAO Xingliu, SHAO Dongsheng, WANG Zhong
    Computer Engineering. 2019, 45(11): 172-176. https://doi.org/10.19678/j.issn.1000-3428.0053016
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Dimensionality reduction and feature selection methods based on single-label classification cannot be directly applied to multi-label learning.If a multi-label learning problem is composed into multiple independent single-label learning problems to enable dimensionality reduction,the correlation between labels will be lost.To address the problem,this paper proposes a multi-label classification algorithm based on embedded feature extraction.The algorithm introduces non-negative matrix factorization into multi-label learning.In feature extraction of original multi-label datasets,the redundant or irrelevant features are reduced and effects of high-dimensional features on multi-label classification are relieved.The comparison experiments are performed on four public standard data sets,and results show that the proposed algorithm can effectively reduce dimensions of data,and lead to better classification performance than traditional BR,CC,LM and other algorithms in terms of multiple evaluation indicators,including accuracy,precision and F measure.
  • KANG Yan, LI Jinyuan, YANG Qiyue, CUI Guorong, WANG Peiyao
    Computer Engineering. 2019, 45(11): 177-182. https://doi.org/10.19678/j.issn.1000-3428.0052829
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Text classification method based on space vector model has high latitude and sparse features in text expression,which leads to poor performance in feature description,and feature engineering relies on manual extraction,the cost of which is high.To address these problems,this paper proposes a text classification algorithm using convolutional capsule network based on dual-channel word vectors.This algorithm uses word vectors trained by Word2Vec and context vectors extended based on specific text classification tasks as two input channels of the neural network.Then a convolutional capsule network model with dynamic routing mechanism is used for text classification.Experimental results on multiple English datasets show that the dual-channel training method for word vectors has better performance than the single-channel training method.Also,the proposed algorithm has a higher accuracy rate in text classification compared with LSTM,RAE,MV-RNN and other algorithms.
  • ZHONG Jian, XU Yang, CHEN Shuwei, HE Xingxing
    Computer Engineering. 2019, 45(11): 183-190,197. https://doi.org/10.19678/j.issn.1000-3428.0054368
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the complex term structure and difficult extraction of grammatical and semantic features in first-order logic,this paper analyzes the constraints and measurement rules of Herbrand semantic features of a term in text replacement.On this basis,the paper gives a definition of the stability of a term,and proposes a stability-based term evaluation method.As a heuristic strategy for text selection,this method is applied to the redundancy judgement of the clause set in Automated Theorem Prover(ATP).Results show that this method can better characterize the features of a term in first-order logic.Compared with the word selection method based on term order,its detection times are reduced by 50.8% on average,and its running time is shortened by 53.3% on average.
  • KE Yan, LIN Xiaozhu, LIAO Rui, WEI Zhanhong
    Computer Engineering. 2019, 45(11): 191-197. https://doi.org/10.19678/j.issn.1000-3428.0053044
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    With the continuous development of deep learning,Convolutional Neural Network(CNN) have received extensive attention from researchers in target detection and image classification.CNN have evolved from LeNet-5 networks to deep residual networks,and the number of layers are increasing.According to the definition of the depth of a neural network,this paper trains CNN of different layers with given standard input images and an output feature graph.The training is performed on the premise that the CNN shares the same receptive field,and training results are compared with standard output graphs.On this basis,the standard 3×3 convolution kernel is decomposed,and a convolutional neural network composed of 2×2 convolution kernels is constructed.According to the central symmetry property of the target feature,the selection rule of the initial weight of the multi-layer convolutional network is proposed.
  • WEI Bingru, ZHANG Guofu, SU Zhaopin, YUE Feng, NIU Fuqiang
    Computer Engineering. 2019, 45(11): 198-203. https://doi.org/10.19678/j.issn.1000-3428.0052532
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The solution space complexity of Overlapping Coalition Structure Generation(OCSG) is high.The stochastic search method based on evolutionary computation can not guarantee the optimal solution,and it assumes that the Agent does not incur any cost when it undertakes the task to consume resources,which makes it impossible to distinguish the differences among coalition structures.To solve this problem,an OCSG mathematical model with the objective of minimizing the cost of coalition structure is constructed,and an optimal OCSG algorithm based on dynamic planning is proposed.Experimental results show that compared with TTGs_DP algorithm,this algorithm has better environmental adaptability and higher resource utilization rate.
  • YIN Ying, JI Lixin, CHENG Xiaotao, HUANG Ruiyang, LIU Zhengming
    Computer Engineering. 2019, 45(11): 204-212. https://doi.org/10.19678/j.issn.1000-3428.0052645
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Currently,most studies on information network are based on homogenous networks,while the studies on network representation learning of heterogeneous networks are relatively rare.Therefore,we propose a network representation learning algorithm,which transforms heterogeneous information networks into weighted homomorphic subgraphs according to different meta-paths.First,we construct weighted edges base on different meta-paths between similar nodes,so as to extract the weighted homogenous subgraphs from heterogeneous network.Then,we obtain the similar node sequence by random walk with offset.Finally,we use Skimp-gram model to generate the representation vector of the nodes.Experimental results show that compared with other algorithms which only consider a single meta-path,the proposed algorithm has better effects on data mining tasks such as node classification and similarity search.
  • LIU Yiting, LI Peifeng
    Computer Engineering. 2019, 45(11): 213-217,224. https://doi.org/10.19678/j.issn.1000-3428.0051052
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The current neural network-based event trigger extraction model distinguishes triggers by entity information,but lots of unrelated entities have negative interference on trigger extraction.Therefore,we propose an event trigger extraction method based on local entity characteristics.First,we filter unrelated entities,separate the reserved entities into core and non-core types,and build the reserved entity model.Then,with the utilization of CNN network,which is superior in the extraction of local features,the model manages to locate the entities that are important for trigger recognition from all the entities.Subsequently,we adopt the Attention mechanism to improve the weights of important entities,so as to identify triggers by their characteristic information.Experimental results on domain-specific and domain-general event corpus show that the proposed method can reduce the interference of unrelated entities on trigger extraction,and its F1-score of trigger extraction performance,if at the highest level,can exceed the baseline by 0.017.
  • GU Xinglong, XIE Jun, JIN Hongwei, XU Xinying
    Computer Engineering. 2019, 45(11): 218-224. https://doi.org/10.19678/j.issn.1000-3428.0052461
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The fine-grained sentiment analysis of short texts in online comment is a research hotspot of text mining.As the basis of fine-grained sentiment analysis,evaluation objects play an important role in the process of identifying texts.How to make full use of context information and effectively represent is the object of evaluation difficulty.Therefore,combining the word feature and semantic feature,an evaluation object identification method is proposed.According to the commodity review corpus,the Conditional Random Fields(CRFs) are used to identify the object of comment,and the semantic feature is introduced on the basis of word feature and dependent syntax feature.Different features are combined to make full use of context information and improve the recognition accuracy of the comment object.Experiments on two data sets of mobile phone reviews and hotel reviews show that the accuracy of the method is higher,and the F values are 75.36% and 82.64%respectively.
  • XIE Zongfu, GAO Bo, YUE Chunsheng
    Computer Engineering. 2019, 45(11): 225-233,242. https://doi.org/10.19678/j.issn.1000-3428.0052218
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In order to realize the fast dynamic loading,unloading and intelligent switching for application tasks in the signal processing platform,an intelligent loading and unloading algorithm based on Ordered Binary Decision Diagram (OBDD) is proposed.Based on the analysis of the whole intelligent loading and unloading system,the software and hardware resource model for the signal processing platform,situation parameter definition dictionary and intelligent loading and unloading rule database of signal processing platform are constructed.Through the intelligent rule decision algorithm based on OBDD,the intelligent loading and unloading for the application tasks under the minimum manual intervention is realized.Experimental results show that compared with the sequential rule storage algorithm,the proposed algorithm has higher efficiency and lower cyclomatic complexity.
  • HU Hangyu, ZHAI Xuemeng, HU Guangmin
    Computer Engineering. 2019, 45(11): 234-242. https://doi.org/10.19678/j.issn.1000-3428.0052745
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The graph model method has unique advantages in network flow behavior analysis,because it can intuitively and completely describe the connection mode of network flow.However,the current methods have many problems,such as single composition mode,incomplete information and insufficient analysis means etc.Therefore,by referring to the concept of knowledge graph,this paper proposes a network flow behavior analysis model based on flow knowledge graph,namely,the network flow connection graph.We first build the basic model of the network flow connection relationship by collecting the network flow information.Then we set the graph node level and the edge weight value based on the network flow attribute information.According to the filtering rules of the node and edge,we extract the core connection mode of the network application behavior and simplify the network scale.Finally,we adopt the complex network feature analysis method to extract the network flow feature parameters.Experimental results show that network flow connection graph can fully utilize the available information in the network flow behavior measurement data,accurately characterize the inherent characteristics of the network application flow connection relationship,and effectively detect and identify network abnormal behaviors such as DDoS attacks,worm propagation and port scanning.Besides,the network flow connection graph shows good scalability,making it suitable for the application of multiple graph mining algorithms.
  • LIU Xueping, LI Yuqian, LIU Li, WANG Zhe, LIU Yu
    Computer Engineering. 2019, 45(11): 243-248. https://doi.org/10.19678/j.issn.1000-3428.0052861
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In order to accurately identify the target parts in the industrial image,an improved YOLOV3 target recognition algorithm is proposed.The K-means clustering and particle swarm optimization algorithm are combined to calculate the anchor box to reduce the influence of the initial point on the clustering result and speed up the convergence of the algorithm.The SE-YOLOV3 network is obtained by embedding the SENet structure after the shortcut layers.A sample set containing 10 816 images is created by collecting the part images and enhancing the data while labeling the part in the image,which is used to train and test the network.Experimental results show that the proposed algorithm can obtain an anchor box with an average IoU of 83.01%.When there are more defective parts in the sample image,YOLOV3 might identify the background as a part,and the precision and recall rate are 72.11% and 97.51% respectively,while the SE-YOLOV3 can accurately identify target parts,whose precision and recall rate are 90.39% and 93.25% respectively.
  • CHEN Chen, ZHOU Yongjun, LI Yuanxiang, TUO Hongya, ZHOU Yu, LUO Jianhua
    Computer Engineering. 2019, 45(11): 249-255. https://doi.org/10.19678/j.issn.1000-3428.0052166
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Image registration is required for the aerial remote sensing images when it is utilized for land mapping and change detection.In order to achieve high-precision matching of target regions,a remote sensing image registration method is proposed.To suit for the small sample data sets,the images are segmented by U-net.Considering the errors of different regional features,Heteroscedastic Errors-in-Variables(HEIV) model is applied to the registration parameter estimation to improve the registration accuracy of the target region.Experimental results show that compared with the Harris corner-based registration method,the global average registration accuracy of the proposed method is improved by 41.39%,and compared with the Scale-Invariant Feature Transform(SIFT) feature-based registration method,the average registration accuracy of the regions of interest is increased by 16.67%.
  • WEI Zhen, JIANG Zhijun, YANG Xiaohui, ZHANG Hao
    Computer Engineering. 2019, 45(11): 256-261,268. https://doi.org/10.19678/j.issn.1000-3428.0052906
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    When tracking targets are subject to distortion,occlusion,fast and irregular motion,correlation filter tracking algorithms based on single color feature are difficult to achieve accurate target localization.To this end,the Kernel Correlation Filter algorithm(KCF) based on multi-channel color feature of Color Names(CN) is analyzed,and an improved algorithm is proposed integrating CN features and color statistical features.The training samples of the CN features are cut by a mask matrix to increase the proportion of real samples.On this basis,the CN features and the color statistical features are used to train the position-dependent filter to obtain the target positions respectively.And then the two positions are weighted to obtain the final target tracking result.Experimental results show that,compared with KCF and the algorithms with better performance in benchmark_tracker database,the tracking precision and success rate of the proposed algorithm are higher under the inference of distortion and occlusion.
  • ZENG Mengqi, MA Weiyin, LI Li
    Computer Engineering. 2019, 45(11): 262-268. https://doi.org/10.19678/j.issn.1000-3428.0052836
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Fusion of visual and text information can bridge the semantic gap between low-level visual features of images and high-level semantics,and thus improve retrieval quality. However,retrieval efficiency is consequently reduced.To address this problem,this paper designs a novel index structure,CAT-tree,and a corresponding top-k retrieval algorithm for hybrid image retrieval based on fusion of text and content.The CAT tree and its top-k algorithm are designed by combining Manhattan Hashing,Inverted Index and R-tree.On this basis,a three-way image retrieval scheme is proposed.Experimental results on benchmark image datasets show that the proposed scheme can remarkably improve the image retrieval efficiency without the loss of accuracy.
  • HU Pan, QIAO Lin, XU Libo, YU Yuanqi, HAN Yonghui
    Computer Engineering. 2019, 45(11): 269-274. https://doi.org/10.19678/j.issn.1000-3428.0052854
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    Currently,most tracking algorithms use scale-traversing poor search strategy to cope with the scale changes of targets,but the tracking performance and efficiency are not satisfying.Therefore,we propose an adaptive tracking algorithm base on specific object proposal.First,we adopt the scale and position information of the tracking target into the object proposal generation algorithm,so as to obtain the specific object proposal and its features.Then,to ensure the tracking continuity,we use the adaptive Support Vector Machine(SVM) as tracking model to score the specific object proposal,thus obtaining the target.Finally,we update the model by classifying the positive and negative samples of specific object proposal and the uniformly-sampled samples.Comparative experiments on OTB100 dataset show that compared with CNN-SVM and DeepSRDCF algorithms,the proposed algorithm can better adapt to the scale changes and deformation of the target,and improve tracking efficiency.
  • LIU Chang, XIE Wenjun, ZHANG Peng, GUO Qing, GAO Chao
    Computer Engineering. 2019, 45(11): 275-280. https://doi.org/10.19678/j.issn.1000-3428.0052945
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    There may be multiple sudden threats in multi-target group,multi-base and multi-UAV collaborative mission planning environment.To address this problem,a joint algorithm of Periodic Fast Search Genetic Algorithm(PFSGA) and Artificial Potential Field(APF) method is proposed.Taking the reconnaissance mission as the background,the common assignment strategy is introduced into the mission planning,the multi-base multi-UAV cooperative mission planning model is constructed,and the PFSGA algorithm is used for initial mission planning.On this basis,sudden threats between bases and target groups are considered,and the APF method is used to avoid obstacles.Simulation results show that the proposed algorithm has good obstacle avoidance performance,and compared with joint algorithm of genetic algorithm and APF,the joint algorithm of PFSGA-APF can avoid falling into the local optimum and obtain the optimal solution easily.
  • TAN Yanghong, ZHANG Jing, LI Xiao
    Computer Engineering. 2019, 45(11): 281-286,297. https://doi.org/10.19678/j.issn.1000-3428.0053015
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The important nodes in the power grid have an important impact on the safe and orderly operation of the power grid.If these nodes fail,it will cause a large area of the power grid.Traditional methods of assessing the importance of grid nodes often ignore the influence of decision makers' experience and knowledge on the results of the assessment.Therefore,the prospect theory is introduced and an improved Technique for Order Preference by Similarity to Ideal Solution(TOPSIS)method is proposed.The introduction of relative entropy solves the problem that the traditional TOPSIS method can not distinguish the vertical line nodes in the positive and negative ideal solutions,and uses the gray correlation degree to measure the Euclidean distance and the edge coupling degree of the curve,so that the evaluation result is more accurate.Taking the IEEE 39 node system as an example,the analysis results show that the method can effectively identify important nodes and the node recognition is high.
  • FU Zhiyue, ZHANG Rui, LI Fufeng
    Computer Engineering. 2019, 45(11): 287-297. https://doi.org/10.19678/j.issn.1000-3428.0052471
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The traditional pulse signal time domain feature point recognition method does not consider the diversity of pulse morphology,and lacks the connection with the pulse wave formation mechanism,thus limiting its recognition accuracy.To this end,the wavelet transform modulus maxima curve and its singularity detection theory are introduced.Combined the pulse wave formation mechanism with the pulse feature point interpretation,a pulse time domain feature point recognition method is proposed.The wavelet transform modulus maxima curve is used to detect the singularities in the pulse signal.According to the property and position relationship of the singularities,the position range of each time domain feature point is determined.The difference method is used to determine the position of each time domain feature point.Experimental results show that the proposed method can not only avoid the interference of pulse frequency instability on feature point recognition,but also adapt to the diversity of pulse morphology.The recognition accuracy of time domain feature points is better than sliding window method,Shannon envelope method and empirical division method.
  • SONG Ningjia, CUI Yinghua
    Computer Engineering. 2019, 45(11): 298-302,308. https://doi.org/10.19678/j.issn.1000-3428.0052838
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    To address the problem that the positioning algorithm based on distance measuring is vulnerable to environmental interference and large measurement errors,a fingerprint positioning algorithm based on Genetic Algorithm-Generalized Regression Neural Networks(GA-GRNN) is proposed.The GRNN is used to establish the node positioning model.The GA is used to determine the optimal smoothing parameters of the generalized neural network.The signal strength value between the reader and the tag is used as the input of the neural network,and the output node coordinates are obtained.The simulation results show that compared with GRNN algorithm,BP neural network algorithm and FOA-GRNN algorithm,the algorithm has higher positioning accuracy and strong generalization ability.
  • WANG Qianzhu, WEI Qingxia, YANG Xiaoya, KANG Lulu
    Computer Engineering. 2019, 45(11): 303-308. https://doi.org/10.19678/j.issn.1000-3428.0052925
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In order to accurately evaluate the user's car purchase tendency,an evaluation method on car purchase tendency based on the subjective and objective combination weighting is proposed.The compatibility of multi-weighting method is used to ensure the rationality of index weights and reduce the number of combined weights.According to the correlation characteristics between the car purchase tendency and the purchased car event,the real car user data is used to construct the ideal point,and the index weight is corrected to improve the car purchase tendency evaluation performance.Experimental results show that compared with traditional ideal point method and unscreened direct weighting method,the computational complexity of the method is reduced and the accuracy of the evaluation effect is improved.
  • ZHOU Siyu, LI Xuejun, XU Jia, YE Chunsen
    Computer Engineering. 2019, 45(11): 309-314,320. https://doi.org/10.19678/j.issn.1000-3428.0052832
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    In supply chain management,the uncertainty of customer demand will cause bullwhip effect,which can increase the deviation of demand information.However,the most applied supply chain strategy is vendor managed inventory,which fails to consider the influence of uncertainty.To address this problem,we propose a joint replenishment strategy for inventory management based on robust optimization,which is specifically for the uncertain demands of single supplier and multiple retailers.Firstly,we build a non-linear hybrid integer programming model to calculate the total cost of two-stage supply chain,so as to reflect the performance of the supply chain system through the changes of total cost.Then,we use robust optimization method to get the minimum total cost of the supply chain system.Finally,we use the external and internal iterative algorithms to obtain the replenishment period and replenishment number of suppliers and retailers.Experimental results show that compared with the traditional supply chain strategy ERI and AR,the proposed strategy can effectively reduce the total cost of the supply chain system.
  • WU Mengnan, LI Lihong
    Computer Engineering. 2019, 45(11): 315-320. https://doi.org/10.19678/j.issn.1000-3428.0052375
    Abstract ( ) Download PDF ( ) HTML ( )   Knowledge map   Save
    The 3D point cloud model of vehicle contour plays a key role in the intelligent manufacturing and maintenance of automobiles.In order to improve the accuracy and efficiency of point cloud registration,this paper taking the vehicle maintenance robot as the research object,proposes a vehicle outline scanning and positioning method based on point cloud data processing technology,as well as a date registration method for point cloud.Install a Kinect depth sensor at the end of the arm for precise movement.The point cloud data is collected and pre-processed around the car.The sensor sample pose is calculated according to the kinematic equation of the manipulator to complete the preliminary registration.On this basis,the iterative nearest point algorithm is used to implement precise point cloud registration for vehicle outline.Experimental results show that the method can complete the accurate and fast registration of cloud data from various viewpoints and obtain a complete 3D point cloud digital model.