15 May 2018, Volume 44 Issue 5
-
Select all|
-
Computer Engineering. 2018, 44(5): 1-6. https://doi.org/10.19678/j.issn.1000-3428.0046289
The analysis and query of trajectory data have very important application value in the fields of mobile data management,location services and so on.In order to improve the analysis and mining efficiency of massive multivariate trajectory data,a new multiplex trajectory data index method MTSAX is proposed.A multi-dimensional spatial coding method is given:GeoWord coding,based on the iSAX index framework,designing a moving object track index method.Experimental results on real trajectory datasets show that MTSAX can achieve better trajectory query performance compared with the traditional benchmark methods.
-
Computer Engineering. 2018, 44(5): 7-13. https://doi.org/10.19678/j.issn.1000-3428.0046728
The existing bursty event detection method does not consider the importance of the eveuts,and treats the bursty event time domain and spatial domain of the incident in an isolated manner,and proposes an incident detection method based on comprehensive analysis of spatio-temporal elements.The data cube structure is introduced to store event words,and important events are extracted by a real-time event clustering algorithm based on semantic similarity.TFIDF is used to calculate the occurrence weights of events in the space-time dimension,and the finite state machine-Gaussian distribution model is used to identify spatio-temporal events.Experimental results show that the method can effectively identify bursty time and bursty area of the event,compared with the existing emergency detection method,the accuracy of detecting eveuts is higher.
-
Computer Engineering. 2018, 44(5): 14-18,24. https://doi.org/10.19678/j.issn.1000-3428.0046777Combined with multi-level Haar wavelet transform and KS statistic theory, this paper proposes a fast detection method for time series data abrupt-point,that names as HWKS.The mean binary search tree and the difference binary search tree are constructed respectively for the standard reference sequence and the sequence to be detected.Based on the improved KS test method,two methods of binary tree search are proposed,and the HWKS theory framework for rapid detection of time series data mutation point is realized.HWKS is used to detect the simulated time series data.Comparison results on HW method,T method and KS method show that HWKS has less error,shorter time and higher accuracy when detecting time series data abrupt-point.
-
Computer Engineering. 2018, 44(5): 19-24. https://doi.org/10.19678/j.issn.1000-3428.0046453The training time of Support Vector Machine(SVM) becomes much longer with the increase of training instances.To speeding up the training process without loosing the effect,a method for reducing training dataset is proposed.A new training set is composed with nearest neighbors of each instance selected from different classes.A boundary sample is selected by using an alien nearest neighbor.Experimental results show that compared with FCNN algorithm and NPPS algorithm,the proposed algorithm can enormously raising the training efficiency while keeping or even improving the classification accuracy of SVM.
-
Computer Engineering. 2018, 44(5): 25-32,46. https://doi.org/10.19678/j.issn.1000-3428.0046381Aiming at the problem that most traditional clustering algorithms cannot cluster time-series data flow,a clustering algorithm based on density and grid is proposed.It introduces the method of dynamic grid partitioning,dynamically set parameters such as grid partition and grid density thresholds and adaptively generates grids through the characteristics of data within the current data block,convert it into different types of maps and performs clustering separately.The trajectory of an individual is analyzed,and the individual trajectory is detected by the individual trajectory division method according to the time period.Experimental results show that the algorithm can obtain the clustering results of data in different time periods according to the user’s requirements,and is suitable for the problem of abnormal detection of time-series data flow such as trajectory information.
-
Computer Engineering. 2018, 44(5): 33-39. https://doi.org/10.19678/j.issn.1000-3428.0046594In view of the problems of intercommunication between different communication devices and difficulties of software and hardware upgrading,with the purpose of improving the portability and sharing of components,according to the architecture of new heterogeneous platform,a reconfigurable component for signal processing platform is designed.Based on the characteristic of reconfigurable components in GNU Radio platform,combined with the advantages of software reconstruction technology,text engine,abstraction layer technology and message middleware,a design method of reconfigurable components based on heterogeneous signal processing platform is proposed.Considering a sample of envelope squares estimation algorithm,the flow graph of the signal processing components is written in GNU Radio and heterogeneous signal processing platform.Experimental results show that the mean square error of the parameters estimated by the reconfigurable component is consistent with the data in Matlab tool,and has transplantability across the platform.
-
Computer Engineering. 2018, 44(5): 40-46. https://doi.org/10.19678/j.issn.1000-3428.0046331It takes long time to recover when a single point of failure occurs in Redis cluster and the recover process is very unstable.This paper improves the communication process of the fault tolerance system in Redis cluster based on the consideration of load balancing without increasing the transmission cost of the system,on the purpose of improving the reliability of the cluster.Experimental results show that this method has a great improvement on the reliability and failover effectiveness of Redis cluster,under the premise that it does not cause much impact on the network transmission cost and throughput of the cluster,and indirectly enhances the reliability of cluster processing fault state.
-
Computer Engineering. 2018, 44(5): 47-52,59. https://doi.org/10.19678/j.issn.1000-3428.0047957In order to make full use of the advantages of embedded multi-core ARM microprocessor,such as small size,low power consumption,low cost and high performance to improve the system speed,this paper studies the architecture of ARM and the optimization problem of basic function based on the architecture.It analyzes the string and memory processing functions in Bionic library based on the structural characteristics of ARM v8 on the ARM Cortex-A72 platform.Experimental results show that after program level optimization of whole word processing,loop unrolling and special instruction optimization techniques,the performance of commonly used base functions in the Bionic library is improved to varying degrees on the ARM Cortex-A72 platform.
-
Computer Engineering. 2018, 44(5): 53-59. https://doi.org/10.19678/j.issn.1000-3428.0047130The overlay network tool,Flannel,used by Kubernetes implements the container service communication through NAT,which will produce large network performance loss.Therefore,based on the Calico network container application platform,a kind of application scheme for micro-service container is proposed.The performance evaluation of several virtualized network software is carried out by using the network performance testing tool IPerf3.Combining the features and functions of the network software with the Kubernetes integration,the application platform of micro-service containerization is realized.Experimental results show that,in terms of throughput and transmission rate,the proposed scheme can improve the network performance by about 50% compared to Flannel,which is more conducive to the application of micro-service in the actual production environment.
-
Computer Engineering. 2018, 44(5): 60-65,77. https://doi.org/10.19678/j.issn.1000-3428.0046855Time verification is inadequate in workflows of Mission Critical System(MCS).To solve this problem,a new workflow modeling and time verification approach is proposed.Firstly,this approach divides the system tasks into mission critical ones and non-mission critical ones based on Place Containing the Time factor colored Petri-Workflow Net(PCTP-WFN),this can manage the activities which cannot meet the time constraint,and can execute non-mission critical tasks as many as possible on the premise of finishing the final tasks,which can thereby eliminate much wasted time in the MCS.Secondly,this approach transforms the time constraint of workflow activities into mathematical constraint,then uses MATLAB to analyze the models,and gets the optimal path finally.The example analysis of the shore-based facility security information system show that,the proposed approach is feasible and practicable.
-
Computer Engineering. 2018, 44(5): 66-70. https://doi.org/10.19678/j.issn.1000-3428.0046933The existing design of the MavLink protocol is mainly aimed at low-rate,large-data-rate message transmission,there are problems such as poor characteristics of mobile communication channels,and many environmental interference factors.Therefore,a new type of Mini Unmanned Aerial Vehicle-Mobile Transfer Protocol(MUAV-MTP) protocol is proposed for mobile transmission authentication,routing and control information.By referring to the existing mobile MavLink communication protocol,the original protocol is converted into a remote mobile control protocol.USA’s cruising movement control is achieved through a short instruction packet,and the TCP and UDP protocols are supported at the same time.Experimental results show that the MUAV-MTP protocol has the advantages of simple operation,universal interface,good extensibility and robustness,and can be used to test and develop UAV based on Pixhawk flight control.
-
Computer Engineering. 2018, 44(5): 71-77. https://doi.org/10.19678/j.issn.1000-3428.0046206In the Wireless Sensor Network(WSN),the whole coverage of the three-dimensional target under the three-dimensional space needs to consider the three-dimensional structure of space target which irregular influence on deployment,and how to avoid the collision between the nodes on the premise of keeping to maximize the target surface area of coverage.Aiming at these problems,this paper presents an object self-coverage algorithm in three-dimensional space.The method of three-dimensional centroid calculation based on the convex set determines the scope of the target center and node deployment,and the uneven virtual potential field method is used to overcome the overlapping and covering hole problems,which translates mobile sensor network deployment process into the virtual potential field’s nodes self-mobile and self-rotation process by the virtual force.Simulation result shows that,compared with the spatial partitioning algorithm,the proposed algorithm has better performance on coverage and energy consumption.
-
Computer Engineering. 2018, 44(5): 78-82. https://doi.org/10.19678/j.issn.1000-3428.0046075
At present,the evaluation methods of mobile application mostly focus on the association mining of the vulnerability collection and malicious sample behaviors,which is difficult to cope with the rapid increase in mobile applications.Aiming at this problem,based on the attack graph,this paper takes security loopholes and attack threats as node and establishes a mobile application security threat assessment model.It makes the mobile application vulnerabilities and attack threats interconnected based on the relevance.At the same time,it considers the quantification of the vulnerability risk value and the impact of the relevancy on the assessment results when evaluating the security threats of mobile application.Experimental results show that this model can improve the accuracy of mobile application security threat assessment,and can be used for large-scale mobile application model building with good scalability.
-
Computer Engineering. 2018, 44(5): 83-87,93. https://doi.org/10.19678/j.issn.1000-3428.0046673In order to minimize the charging cost in wireless sensor networks,two wireless charger location optimization algorithms are designed according to the location and topological features of sensor nodes:approximate algorithm and cluster partitioning algorithm.The segmentation technology is used to divide the network into several grids.The minimum number of chargers in each grid is solved separately.The number of chargers is reduced by the shift strategy,and an approximation ratio approximation (1+ε)algorithm is obtained.At the same time,the greedy thought is adopted to cluster the clusters,and the clustering partitioning algorithm is proposed based on the least enclosing circle algorithm.Complexity analysis and simulation results show that both algorithms can effectively solve the position of the charger and are superior to other algorithms.The clustering partitioning algorithm has fewer number of chargers and lower time complexity.
-
Computer Engineering. 2018, 44(5): 88-93. https://doi.org/10.19678/j.issn.1000-3428.0046425Aiming at the existing Message Passing Algorithm(MPA) based on parallel strategy and serial strategy,and the parallel MAX-Log MPA based on log domain,there is a lack of comparative analysis of algorithm complexity and detection performance,the implementation principle of the above 3 algorithms is expounded.Combined with the advantages of serial MPA algorithm and parallel MAX-Log MPA algorithm,a serial MAX-Log MPA algorithm based on logarithmic domain is proposed.The updated messages are passed to the later nodes to make full use of the new information and speed up the convergence of the algorithm.And the computational complexity is further reduced by the reduction of the number of iterations required for convergence.Simulation results show that,compared with the MPA algorithm based on parallel strategy,the proposed algorithm has fast convergence speed and low algorithm complexity.
-
Computer Engineering. 2018, 44(5): 94-100,106. https://doi.org/10.19678/j.issn.1000-3428.0046140In-network cache is one of the most important features of Named Data Networking(NDN),aiming to solve the efficient utilize of the cached replicas highly,the local popularity based distributed cooperative cache is proposed.Count the rate of the content request of the user access nodes periodically,and the results are forwarded with the interest package,at the result of avoiding the filter effect.Implement cache decisions based on cache the revenue on the request path;between the neighbor nodes are performing local cache announcements which can improve the cache utilization.Simulation results show that the proposed strategy can reduce the request delay,improve the cache hit rate.
-
Computer Engineering. 2018, 44(5): 101-106. https://doi.org/10.19678/j.issn.1000-3428.0046471In order to solve the problem of low physical layer security performance in heterogeneous networks,an In-Band Full-Duplex(IB-FD) wireless backhaul scheme is proposed.In massive Multiple Input Multiple Output (MIMO) two-tier HetNets,macro and small cell links are eavesdropped by malicious eavesdroppers,and the wireless backhaul signal is regarded as the interference signal for macro and small cell communications to enhance communication link quality.By using the tool of stochastic geometry and probability statistics,we modeled the network elements as independent Poission point processes,and investigate the secrecy probability of the macro cell downlinks and small cell uplinks.The presented results show that in massive MIMO HetNets,exploiting the IB-FD wireless backhaul signal as the interference can enhance the secrecy probabilities of both the macro and small cell transmission,and the in-band full-duplex backhaul mode outperforms the out-band Full-Duplex(FD) one.
-
Computer Engineering. 2018, 44(5): 107-112,123. https://doi.org/10.19678/j.issn.1000-3428.0047759In order to realize the effective connectivity of a large number of mobile combat units in the future aeronautic swarm operation,this paper proposes an idea of using Unmanned Aerial Vehicle(UAV) to build a relay network,and researches the deployment of the network.Firstly,according to the characteristics of UAV relay network,the concept of node aggregation degree is proposed.In order to deploy by considering the combat unit coverage,network connectivity and survivability,a performance evaluation parameter for UAV relay network is defined based on node aggregation degree,and a Virtual Force locally Adjusted Particle Swarm Optimization(VFA-PSO) algorithm based on virtual force local adjustment is designed.Experimental results show that the proposed algorithm can effectively improve the coverage rate and survivability of the network,meanwhile getting better deployment effect under the premise of ensuring the network connectivity.
-
Computer Engineering. 2018, 44(5): 113-118. https://doi.org/10.19678/j.issn.1000-3428.0046314The program written by Python programming language is compiled and generated bytecode,which is a specific structure for Python virtual machine.The file is easily decompressed by reverse tools,thereby damaging the economic interests and personal privacy of developers.The traditional method of anti-reversal has the problems of easily deciphered bytecode files and low program efficiency after processing.To solve the above problem,a new Python bytecode file protection algorithm is proposed,which mereges multiple opcodes in the Python bytecode file into a new opcode without changing the execute result of the program,changes the opcode sequence structure and semantics,and achieves the purpose of anti-reversal finally.Experimental results show that the proposed algorithm not only prevents the bytecode files from being decompilated,but also reduces the storage space of the bytecode file and improves the efficiency of program execution.
-
Computer Engineering. 2018, 44(5): 119-123. https://doi.org/10.19678/j.issn.1000-3428.0046506The information in the traditional Radio Frequency Identification(RFID) system during the communication process may be intercepted by malicious users,resulting in the leakage of confidential information.Therefore,a method of improving the security performance of the physical layer of RFID system is proposed.Using security capacity as an evaluation index,two different scenarios are analyzed:When eavesdropper information is known,the security of the communication is ensured through relay selection,which maximizes the security capacity,when the eavesdropper information cannot be known,manual interference is used to reduce the eavesdropper’s ability geting the best power distribution scheme.Simulation results show that this method can greatly improve the system security capacity with minimal interference power.
-
Computer Engineering. 2018, 44(5): 124-127. https://doi.org/10.19678/j.issn.1000-3428.0048189An identity-based anonymous authentication scheme in Vehicle Ad Hoc Network(VANET) is analyzed.It is demonstrated that the attacker can forge authentication proof using the intercepted secret key and public key because of sending the blinded secret key and public key as authentication proof in the open channel.Aiming at this problem,an improved authentication scheme is proposed.It constructs the signature of timestamp using the one-off public key as authentication proof,so as to make the attacker cannot obtain the secret key.Analysis result shows that the improved scheme can resist forgery attack while realizing anonymous authentication.
-
Computer Engineering. 2018, 44(5): 128-132. https://doi.org/10.19678/j.issn.1000-3428.0046456In the cooperative analysis of intrusion information of networks,the sharing of alerts data will bring the risks of privacy leaks to the attacked ones.Considering that the existing methods for protecting the privacy of alerts information cannot be able to deal with malware analysis with arbitrary background knowledge,a new correlation method for network alerts based on differential privacy protection is presented.The origin alert sequences dataset is the input and the Laplace mechanism is used to build the noisy prefix tree which supporting differential privacy protection.On this basis,the sanitized alert sequences dataset is generated by traversing the noisy prefix tree,and the alerts are correlated by mining frequent sequences.Moreover,this method is proved theoretically that it is able to fulfill ε-differentially private.An experiment is conducted with LLDoS1.0 inside traffic dataset,and the result demonstrates that the proposed method can remain available in the process of preserving privacy.
-
Computer Engineering. 2018, 44(5): 133-139. https://doi.org/10.19678/j.issn.1000-3428.0046507Due to the lack of expressive power,flexibility,update and interaction capability of current Digital Rights Management(DRM) languages,a DRM model named SociDRM which based on social network is proposed.On the basis of the distributed logic language Active-DU-Datalog,the SociDRM model is constructed and its rights protection mechanism is explained.Then the methods of use,rights transfer,reuse and fair use of the digital content are explained.Finally,the implementation method of SociDRM model is discussed and a contrastive analysis to other model which based on XML language is illuminated,the results show that the proposed model is more matching with the social network structure in terms of semantic integrity and copyright protection expression.
-
Computer Engineering. 2018, 44(5): 140-145. https://doi.org/10.19678/j.issn.1000-3428.0046782In order to solve the compounding problem of broadcast audience viewing interest,this paper proposes a TV group detection strategy based on context awareness.Using tensor decomposition to obtain the hidden feature matrix of programs and viewing time,using Markov clustering algorithm to achieve the classification of records,and find the user group according to the record classification results.The user group identifies the specific user as the unit of home users.All home users have similar viewing interests during the time period,and implement program recommendation functions for home user groups.Experimental results show that the strategy reduces the average absolute error in the viewing interests of users and groups within the group,and improves the similarity of viewing interests among members of the group.
-
Computer Engineering. 2018, 44(5): 146-154. https://doi.org/10.19678/j.issn.1000-3428.0048549As the existing nodes and links of local areas are overweight,the phenomenon of packet jitter and resource waste generates when packets transmits between nodes and adjacent links.A virtual network embedding algorithm is designed for load balancing of the whole network in this paper.It considers the differences of nodes and adjacent links resources and put the differences into the node embedding process.The k shortest path algorithm of the adjacency matrix is optimized.The matrix is transformed into the new matrix which could reflect link load balancing.Finally,it analyzes the bottleneck problem of virtual network embedding by dynamically adjusting the resources of nodes and links.Experimental results show that the method could achieve better virtual network embedding ratio and network load balancing than randomized algorithm and greedy algorithm.
-
Computer Engineering. 2018, 44(5): 155-161. https://doi.org/10.19678/j.issn.1000-3428.0048934Aiming at the semantic relationship problem between the expression to be searched and the target document in the existing mathematical expression retrieval system,based on the analysis of the LaTeX expression using the Serialized Feature Extraction(SFE) extraction method,a mathematical expression retrieval method based on ontology is proposed.It use ontology to establish the relationship between mathematical expressions and their concepts and build a mathematical expression semantic ontology library to achieve the purpose of retrieving the semantic related documents of mathematical expressions by inputting keywords,concepts,phrases and mathematical nouns.Experimental results show that the ontology-based mathematical expression retrieval extends the query result set using ontology concepts,so that the recall rate,precision rate,and expansion rate are all improved to some extent.
-
Computer Engineering. 2018, 44(5): 162-167,173. https://doi.org/10.19678/j.issn.1000-3428.0046263
Because of rough classification and missing tags,the existing methods can’t be directly applied to the similar neighbors selection of the mobile users.In order to solve this problem,an efficient similarneighbors selection method for mobile user is proposed.Animproved short text calculation method is based on the mobile services introduction text to find the similarity,thenthe service similarity is incorporated into the method of mobile user similarity calculation,and the unsymmetrical and directed similarity between users is obtained.At last a step screening method is given to solve the problem of the large number of similar neighbors of the target mobile users.Experimental results show that the overall accuracy of the this method is higher than that of the common keyword overlapping method and cosine similarity method.
-
Computer Engineering. 2018, 44(5): 168-173. https://doi.org/10.19678/j.issn.1000-3428.0046862Traditional sentiment lexicon construction methods have problems such as relying on semantic knowledge base,limited coverage and poor domain adaptability.Aiming at these problems,this paper proposes a method to construct sentiment lexicon based on label propagation.The method firstly selects some sentiment seed words manually,and then uses Word2Vec to train word embeddings on corpus and treats the words which have high similarities with seed words as candidate sentiment words,the method also finds out the words which have conjunctive relations with seed words,and treats them as candidate sentiment words.By constructing a semantic association graph through the similarities between seed words and candidate sentiment words,the method uses label propagation algorithm to identify polarities of candidate sentiment words and construct sentiment lexicon.Experimental results show that the method can obtain higher accuracy and better robustness compared with baseline method.
-
Computer Engineering. 2018, 44(5): 174-181,187. https://doi.org/10.19678/j.issn.1000-3428.0048085The structure analysis of social networks is generally implemented through the detection of communities,but the nodes in the real network are hierarchical,a small number of nodes have a large number of connections,so that the network often does not have an optimal division.Therefore,an improved network decompostion method based on removal of important nodes is proposed.The network decomposition strategy based on the removal of the central node is used to analyze the complex network structure,the node centrality is calculated by the fast calculation method of the node involution,the central node is deleted in an iterative manner to decompose the network,and the central node is adjacent to the decomposed branches,the reordering of positions in the matrix yields a hierarchical structure of social networks.Experimental results on a real network show that deleting the node with the highest number of network can quickly decompose the network,according to the network decomposition results,the original network is reduced to a star type with the center node as the core and the decomposed larger branch as leaves as the decomposition,structures can better represent the hierarchical relationships of nodes in the network.
-
Computer Engineering. 2018, 44(5): 182-187. https://doi.org/10.19678/j.issn.1000-3428.0046274The method based on Convolutional Neural Network(CNN) is widely used in the task of sentiment classification.This method mainly uses word vectors as the input of the network.However,in the convolution process,each word vector can only represent one word and does not contain context information,which reduces the continuity of information transmission.Aiming at this problem,a CNN model based on word adjacent feature is proposed,which allows each word vector to carry the characteristics of its adjacent words in the convolution process,which not only guarantees continuity of information transmission,but also ensures the sequentiality of word vectors in local range.The experimental results show that the accuracy rate of the proposed method is 89.43% and 85.61% respectively in the COAE2014 (2 classification) and COAE2015 (3 classification) emotional classification tasks,which verifies the feasibility and efficiency of the model.
-
Computer Engineering. 2018, 44(5): 188-193,200. https://doi.org/10.19678/j.issn.1000-3428.0046702In order to solve the problem of high time complexity and low impact coverage of nodes in influence maximization algorithm,a new influence maximization algorithm is proposed.The PageRank algorithm is used to select the nodes with higher influence as the standby seeds.Then,the combined impact probability of the standby nodes is calculated by counting the number of active rotations and activations of the nodes which can be activated,and k nodes with the largest probability of the combined impact are selected as the seed nodes by the Genetic Algorithm(GA).Simulation results show that compared with DegreeDiscount,PageRank and other algorithms,the algorithm can obtain better nodes selection effect.
-
Computer Engineering. 2018, 44(5): 194-200. https://doi.org/10.19678/j.issn.1000-3428.0046434The form of spam is changeable,easy to disguise as a normal mail and bypass the test,and the new spam has higher detection rate.To solve this problem,based on the idea of negative selection and Support Vector Machine(SVM),a new adaptive classifier is designed and applied to spam detection.The optimal hyperplane of SVM is used to preclassify the mail and get so-called normal emails and spam that match the prediction model,then the second filtration is used in the previous so-called normal emails to get the final normal emails and new spam by the Negative Selection Algorithm(NSA),and the labeled normal emails and spam are used to update the initial optimal hyperplane adaptively,the cycle isn’t stopped until the spam detection rate trend to be stable.The experimental results show that,compared with SVM and NSA,the proposed detection method can improve the recognition rate of new type of spam on the basis of guaranteeing the high recognition rate of normal mail.
-
Computer Engineering. 2018, 44(5): 201-204,214. https://doi.org/10.19678/j.issn.1000-3428.0045763Aiming at the optimization problem of backbone node deployment in Wireless Mesh Network(WMN),a new deployment optimization algorithm of backbone nodes in wireless Mesh networks is proposed.The algorithm adjusts the deployment location of the backbone nodes according to the dynamic changes of user requirements.The TSDPSO algorithm using symmetric displacement mapping adapts to the dynamic environment,detects whether the environment changes at the beginning of a new cycle,and updates the node deployment position to adapt to the change of requirements.On the premise of satisfying user’s bandwidth requirement and network connectivity,Particle Swarm Optimization(PSO) is used to select the gateway location,and the nearest neighbor node with the largest weight is gradually added to minimize the number of routers.Experimental results show that this algorithm can be effectively deployed in dynamic environment.
-
Computer Engineering. 2018, 44(5): 205-208. https://doi.org/10.19678/j.issn.1000-3428.0046937In compressed sensing,the coherence of the measurement matrix directly affects the performance of the reconstruction.Compression ratio is proportional to the reconstruction performance.The requirements for measurement matrix and reconstruction algorithm decrease with the compression ratio.In the case where the compression ratio is less than 0.5,this paper proposes a traverse search method,which is based on orthogonal basis linear representation of deterministic measurement matrices,and breaks the limit of the signal dimension in Hadamard of traverse search.An improved algorithm based on cyclic shift method to construct the orthogonal coefficients is presented to construct a deterministic measurement matrix.Simulation results show that peak signal to noise ratio of the algorithm is higher than that of Gaussian random measurement matrix and Bernoulli random measurement matrix.
-
Computer Engineering. 2018, 44(5): 209-214. https://doi.org/10.19678/j.issn.1000-3428.0046316In order to fulfill the real-time requirement for machine vision system,this paper proposes a new fast image registration method,i.e.Gauss Multi-scale Fast Registration(GMFR).The GMFR algorithm integrates the advantages of registration accuracy and efficiency from Cross Correlation(CC) algorithm in image spatial domain,as well as Cross Power Spectrum(CPS) algorithm in image frequency domain.Based on image multi-scale representation,the image details can be protected via Gauss smoothing during image down sampling.A performance function is further defined in order to perform quantitative analysis of the registration algorithm,and as the image size increases,the efficiency of this new algorithm is improved exponentially.Experimental results on real lead frame images demonstrate that,the proposed GMFR method still has high registration accuracy and strong robustness,even for the complex noise images.Moreover,compared with the traditional CC and CPS image registration methods,the efficiency of the algorithm is significantly improved.
-
Computer Engineering. 2018, 44(5): 215-219,226. https://doi.org/10.19678/j.issn.1000-3428.0046885For improving the poor robustness of pedestrian detectors based on artificial extraction characteristics,learns from the advanced research achievements in the object detection field,proposes a method of real-time pedestrian detection.Being based on the YOLO network structure,combines with characteristics that pedestrians’ aspect ratio is small in images,selects appropriate numbers and scales of the initial candidate boxes by clustering,improves the YOLO network structure,adjusts density of candidate boxes distributed on the X and Y axis,forms a network structure which is suitable for pedestrian detection.Experiments in INRIA pedestrian datasets show that comparing with some pedestrian detection methods,such as HOG,LatSVM-v2 and so on,this method reduces the missed and false detection,improves the positioning accuracy and satisfies the real-time requirements.
-
Computer Engineering. 2018, 44(5): 220-226. https://doi.org/10.19678/j.issn.1000-3428.0046334In the existing parallax image mosaic algorithm,homography matrix are not global,and there is a problem of large computation amount,overlapping ghosting and distortion of the structure.Therefore,a new parallax image mosaic algorithm is proposed.The reference image and the target image are divided into several image blocks with unique properties by using the graph cut algorithm,and the image blocks are numbered.The SIFT algorithm is used to extract the feature of the image,the feature block is added to the feature descriptor to determine the feature matching image block between the target image and the reference image.Through the feature block method,the global homography matrix is calculated and the optimal homography matrix is found to pre-align the target image.The pre-registration image is locally optimized by adding the shape distortion constraint and the block-link constraint in the overlap region to obtain a modified image,and the modified image is fused to obtain the panoramic mosaic image.Experimental results show that the feature block algorithm ensures the global homography,reduces the number of iterations and improves the computational efficiency,the local optimization algorithm can eliminate ghosting and distortion and ensure the quality of image mosaic.
-
Computer Engineering. 2018, 44(5): 227-233,239. https://doi.org/10.19678/j.issn.1000-3428.0048807An improved active contour model is proposed to solve the problem that most of the existing segmentation model can not take into account the segmentation accuracy of the probability noise image and the inhomogeneous image.Firstly,the new neighborhood tensile deformation mechanism is introduced,and the image gradient is used to guide the neighborhood deformation to reduce the effect of the independent information and the noise point on the segmentation results.Secondly,the local correlation coefficient is introduced into the energy function,so that the model can accurately identify the important pixels.Finally,by minimizing the energy function,the active contour is driven to the target edge.Experimental results show that,the proposed model can effectively segment the probability noise with weak boundary properties,and its segmentation accuracy is higher than the adaptive LCV model,LCK model and other models.
-
Computer Engineering. 2018, 44(5): 234-239. https://doi.org/10.19678/j.issn.1000-3428.0046925
Based on the Flexible Representation for Quantum Image(FRQI),a secure quantum watermark protocol is proposed.The new protocol is enhanced in invisibility by making full use of continued-fraction algorithm.It is impossible for anyone except the copyrighter to extract and recover the water image during the protocol execution,which ensures the security of the watermark image,while also providing effective copyright certificate.Simulation results show that compared with the traditional watermark protocol,the size of watermark image reaches the maximum value and the capacity of the new protocol is expanded to eight times that of the previous protocols,and it can better measure the performance of the computation load that is executed by the protocol.
-
Computer Engineering. 2018, 44(5): 240-245. https://doi.org/10.19678/j.issn.1000-3428.0048490To overcome the drawback of bad robustness detection of the existing saliency detection methods,this paper presents a novel saliency detection based on spatial convolutional neural network.The candidate object areas are conformed via preprocessing methods of the removing mean and normalization.By introducing the Convolutional Neural Network(CNN),a global model that captures the context information of saliency objects is constructed,and the corresponding target detection saliency map is obtained.On the other hand,the feature sub-network structure is established to output the 6-dimensional transformation matrix,which is used to transform input image through the spatial deformation module to obtain edge information of salient object.The output of the spatial transformer network local confidence coefficient is introduced into the global information saliency map to seek the maximum value of the feature expression.The discrimination of saliency and non-saliency is realized to accomplish the saliency detection task.Experimental results show that the proposed algorithm improves AUC accuracy under the same condition,generates a saliency map of focus highlighting and achieves impressive robust detection results.
-
Computer Engineering. 2018, 44(5): 246-251. https://doi.org/10.19678/j.issn.1000-3428.0046059In order to improve the accuracy of face recognition and gender recognition in a non-restrictive environment,a facial feature extraction method based on deep convolution neural network is proposed.The VGG-Face model,which is pre-trained by face recognition on a large-scale dataset,is used to fine-tune five of the five attributes on the CelebA face attribute dataset by using a general to special fine-tuning scheme.Face attribute model is gained,and it connects the network full connection layer features as the facial feature vector.Training and testing are conducted on the Adience dataset using a random forest classifier,and high-dimensional data are processed using a random forest approach to select features that are more important for age and gender classification.Experimental results show that this method can overcome the influence of complex illumination and pose changes,and can accurately estimate the age and gender of human face under natural scenes and achieve good effects.
-
Computer Engineering. 2018, 44(5): 252-255. https://doi.org/10.19678/j.issn.1000-3428.0046947Traditional stereo video parallax estimation algorithm based on blocks is difficult to be implemented in hardware,because of its complexity and low efficiency.To solve this problem,parallax pattern recognition algorithm based on three-dimensional self-organizing mapping is proposed.According to the mean values of image blocks,establish pattern library with high brightness and pattern library with low brightness for the parallax sequence.Input blocks can search their matching pattern blocks in different libraries according to mean values.In the library training,the frequency sensitive method is introduced to improve the utilization of vectors,reduce the dead neurons,and decrease the response frequency of the winning vectors.In the experiments,the average Peak Signal-to-Noise Ratio(PSNR) of the algorithm is 1.78 dB,and that results are higher than the traditional parallax estimation algorithm.
-
Computer Engineering. 2018, 44(5): 256-261. https://doi.org/10.19678/j.issn.1000-3428.0046903In order to solve the discontinuous problem of the F0 modeling in the process of the emotional voice conversion,and improve the emotional naturalness of the generated voice,a method for F0 conversion of emotional voice with parameter control based on Non-negative Matrix Factorization(NMF) is proposed.The Continuous Wavelet Transform (CWT) is used to parameterize F0 and model the different levels in the phonetic prosody structure.Then,the characteristic data of F0 is decomposed to the base exemplars and their weights by using the NMF,and by replacing the base exemplars of being converted voice with target,the F0 of the target voice is constructed.In addition,the activation factor,as control parameter,is introduced to optimize the existing model.Experimental results show that,this proposed method has a certain advantage in both the fundamental frequency reconstruction error and the emotional intensity,and can effectively convert the neutral voice to the emotional voice.
-
Computer Engineering. 2018, 44(5): 262-267. https://doi.org/10.19678/j.issn.1000-3428.0046946The performance of speaker aware training method based on i-vector is poor because of using MFCC which has the relative poor robustness as the input feature for the extraction of the i-vector.To solve this problem,an improved i-vector based speaker aware training method is proposed.Firstly,a low dimensional feature extraction method based on SVD is proposed,and then the feature extracted by this method is used to replace the MFCC,which can extract better i-vector.Experimental results show that,in the Vystadial_cz corpus,compared with the DNN-HMM speech recognition system and the original i-vector based speaker aware training method,the recognition performance of this method is increased by 1.62% and 1.52% respectively,in the WSJ corpus,the recognition performance of this method is increased by 3.9% and 1.48% respectively.
-
Computer Engineering. 2018, 44(5): 268-273,279. https://doi.org/10.19678/j.issn.1000-3428.0046615The Streaming Media Edge Cloud(SMEC) based on Software Defined Networking(SDN) needs effective path planning strategy.However,the traditional path planning strategy in SDN only considers the load of network and topology information,there are some defects both in the execution efficiency and optimization effect.To solve the problem,a dynamic hierarchical path planning strategy is proposed.It uses the relative independence of different user regional service requests,and divides the original problem into multiple sub problems by the virtual division of physical network resources,and considers the service status and video popularity information in the process of solving,so as to allocate the network resources.Experimental results show that the proposed strategy can effectively avoid network congestion,improve the service quality of the systems and maintain a low service cost.
-
Computer Engineering. 2018, 44(5): 274-279. https://doi.org/10.19678/j.issn.1000-3428.0046754With the rapid development of multimedia technology tampering software becomes more and more easy to be obtained and used.However,the subsequent forged video interfere the justice of judgment seriously in the meantime.So the detection of video forgery has become essential to enable the integrity of video data.In this paper,a spatial and temporal detection algorithm for detecting object deletion is proposed to mine residual information from the aspects of temporal domain and spatial domain respectively.At the first time,it obtains a good observation view by making use of the video profile to locate the video tampering region,when extracting residual information from temporal domain.Moreover,the algorithm can transforms the 3D video into 2D image effectively.Experimental results show that the proposed approach achieves excellent results on detection.
-
Computer Engineering. 2018, 44(5): 280-285,295. https://doi.org/10.19678/j.issn.1000-3428.0046812In-network caching is one of the key aspects of Named Data Networking(NDN).The existing caching strategy of NDN takes insufficient consideration on caching cooperation among nodes,which can easily lead to the adjacent node to cache the same content and reduce the overall caching efficiency.To solve this problem,this paper builds a cooperative caching model employed in global domain,it is difficult to solve this model,so an algorithm of reservation cooperative cache based on Grey Wolf Optimization(GWO) algorithm is proposed which named RCC.The algorithm uses the sigmoid function to binarize the continuous position update function of the GWO algorithm and uses the position detection and correction algorithm to correct the wolf′s illegal position.Simulation results show that,compared with the CVX optimizer,the RCC algorithm can obtain the approximate optimal solution of the problem with less time and memory,and compared with the traditional caching strategy,the RCC algorithm achieves higher caching hit rate and smaller average hops.
-
Computer Engineering. 2018, 44(5): 286-290. https://doi.org/10.19678/j.issn.1000-3428.0046379At present,most energysaving technologies of smartphone is through restricting user behavior.In order to balance energy saving and user experience,a method of energy consumption optimization for mobile phone is proposed for user behavior.It collects users to use the application’s network activity data when the screen is off.Through mining,classification and induction to generate decision trees,forecasting the importance of application to users,and establishing application network request’s optimal control mathematical model.The optimal control problem is solved by using multi-knapsack algorithm,which restricted the network request of the application after the screen is off.Experimental results show that the energy saving percentage of the method is maintained at about 38.4%,compared with the delayed tolerance,batch processing and other energy-saving technology increased 19.9%,the proposed method can improve the energy consumption problem of smartphone without affecting the user experience.
-
Computer Engineering. 2018, 44(5): 291-295. https://doi.org/10.19678/j.issn.1000-3428.0046769
With the development of information science and technology,the traditional port number and depth packet detection classification technology can not meet the classification requirements of various applications in the network,and can not be classified accurately.A Markov model network traffic classification algorithm based on semi-supervised learning is proposed.The Markov model is constructed by the correlation between flows.The center of the clustering is estimated by density calculation.The center point is calculated by KL distance.The similarity between samples is divided into different application types.The feature of the Markov model is used to identify the traffic application type and improve the accuracy.The problem of the traditional traffic classification method based on semi-supervised learning depends on the unstable clustering algorithm.Experimental results show that the network traffic classifier can achieve the ideal classification effect.
-
Computer Engineering. 2018, 44(5): 296-302,308. https://doi.org/10.19678/j.issn.1000-3428.0047111The existing handoff algorithm lacking of considering status about the next time networks and user status and the user′s demand for transmitting different service when the small-scale aircraft is connected to the aerospace heterogeneous wireless networks.In order to solve the problem,a Q-learning optimized algorithm is proposed to support the heterogeneous wireless networks.The algorithm introduces user experience into reward function,and uses the comprehensive prediction method to get the current and next networks Signal to Interference and Noise Ratio(SINR),user mobility speed,networks handoff cost,information transmission delay and networks congestion degree as comprehensive evaluation parameters.Then,it applies the Analytic Hierarchy Process(AHP) method to determine the evaluation parameter weight according to the service type difference in transmission networks.Simulation results show that,the proposed algorithm has better preference in improving the successful handoff ratio and reducing the handoff number than the traditional algorithm based on Q-learning,when there are different types of services in the networks,the proposed algorithm can provide better networks handoff strategy and reduce the blocking ratio of handoff.
-
Computer Engineering. 2018, 44(5): 303-308. https://doi.org/10.19678/j.issn.1000-3428.0046874Hierarchical heterogeneous networks change the topology of traditional cellular networks,and reduce networks energy consumption on the basis of improving networks capacity.However,the serious interlayer interference will cause significant deterioration of networks energy efficiency.To solve the above problem,this paper proposes Almost Blank Subframe (ABS) ratio dynamic optimization algorithm based on enhanced Inter-Cell Interference Coordination(eICIC) in hierarchical heterogeneous networks to improve networks energy efficiency.The relationship between the networks energy efficiency and ABS ratio is derived firstly.Then by taking the maximum value of the function,ABS ratio is optimized to realize the networks energy efficiency maximization.Simulation results show that,compared with the maximum throughput and the maximum fairness of ABS ratio optimization algorithms,the proposed algorithm can effectively improve the energy efficiency of the networks.
-
Computer Engineering. 2018, 44(5): 309-315. https://doi.org/10.19678/j.issn.1000-3428.0047595In an extended target Gaussian-mixture Probability Hypothesis Density(PHD) filtering,the partition of measurement set requires a lot of calculation,which leads to the decrease of the efficiency of the algorithm.To solve the problem,a new measurement set partitioning algorithm is proposed.Firstly,it uses the Local Outlier Factor (LOF) technique to remove the clutter,then the hierarchical clustering combining with the new method using density peaks is used to partition the measurement set.Simulation results show that compared with distance partition、K-means++ partition、DBSCAN partition,the proposed algorithm is not sensitive to the shape and can ensure the performance of the extended target tracking ,as well as reduces the computational time effectively in the case of target crossover and neighborhood.
-
Computer Engineering. 2018, 44(5): 316-321. https://doi.org/10.19678/j.issn.1000-3428.0046583As DS theory has too little evidence,it may lead to an unstable matching results problem for map matching algorithm.Based on the DS theory,this paper proposes a three evidences DS theory and dual mode map matching algorithm.It uses the direction,distance,and history of three evidences to draw a conclusion,and it exploits linear fitting of timiza tion to demonstrate the conclusion,and outputs projection.For different sections of the road which are adopted the same kind of pattern matching resulting in low efficiency problem,divide the road structure and uses the intersection mode and road section mode respectively to match.Experimental results show that,compared with the other DS theory map matching algorithms,three evidences DS theory map matching algorithm has higher accuracy and matching time is nearly shortened by one-third.
2024
Volume 50 No 4
Volume 50 No 4
(Started in 1975, Monthly)
Sponsored by：
East-China Institute of Computer Technology Shanghai Computer Society
Edited & Published by：
Editorial Board of Computer Engineering
Chief Editor：Jiang Bo
Executive Chief Editor:
Hu Bing
Tel：021-67092217
E-mail: ecice06@ecict.com.cn
CN: 31-1289/TP
ISSN: 1000-3428
Sponsored by：
East-China Institute of Computer Technology Shanghai Computer Society
Edited & Published by：
Editorial Board of Computer Engineering
Chief Editor：Jiang Bo
Executive Chief Editor:
Hu Bing
Tel：021-67092217
E-mail: ecice06@ecict.com.cn
CN: 31-1289/TP
ISSN: 1000-3428