International Journal of Information and Communication Technology (89 papers in press)
LoRaWAN Possible Attacks, Proposed Countmeasures and Enhanced Solution
by Sarra Naoui, Mohamed Houcine Elhdhili, Leila Azouz Saidane
Abstract: The Internet of things (IoT) is pervading our lives to provide a comfortable and smarter human living space by connecting sensors and objects directly with one another without human intervention. The requirements for the deployment of millions of IoT objects are increasing rapidly with a major security concern. To face this challenge, security solutions based on cryptography might be used. In this paper, we investigate one of the widespread IoT security solutions, that is LoRaWAN technology. We evaluate its security by gathering the possible attacks. Afterwards, we propose an enhanced LoRaWAN architecture that faces these attacks. We evaluate the proposed solution in terms of security and performance parameters and compare it to existing solutions.
Keywords: LoRaWAN; Vulnerabilities; Security; Countermeasures.
Fast Road Scenarios Recognition of Intelligent Vehicles by Image Processing
by Huawei Wu, Yicheng Li, Jie Zhang, Yong Kuang
Abstract: Road scenarios recognition is a key point for intelligent vehicles (IVs) self-localization. This paper proposes a fast road scenarios recognition method by image processing, which is based on a visual road model. First, a sequence of images is used to set up the visual road model. Second, we use the classic ORB (Oriented FAST and Rotated BRIEF) method to encode both holistic and local features of a query image. In the step of holistic feature extraction, each query image is treat as a feature and the feature size is only 63
Keywords: intelligent vehicle; computer vision; road scenario recognition; holistic feature; local feature;hybird KNN.
Security Protocol of Internet of Things based on NTRU Encryption Model
by Ying Li
Abstract: In view of the low security and reliability of the current Internet of things security protocols, a NTRU encryption model based Internet of things security protocol method is proposed. This method uses NTRU-PGP to construct the security protocol of the Internet of Things, reduces the amount of data through PKZIP algorithm, introduces AES symmetric encryption algorithm to encrypt the information data that need to be transmitted in the Internet of Things, re-encrypts the encrypted information according to NTRU encryption algorithm, and implements information signature by NTRUSign algorithm to ensure the security of the Internet of Things. The experimental results show that when the proposed method is attacked by active attack, passive attack, eavesdropping data, counterfeiting and other types of attacks, the estimated attack time is over 1025, which is significantly higher than other data, which proves that the protocol is more secure and reliable.
Keywords: NTRU; Internet of Things; Security protocol.
Lightweight Information Integration Model of IOT based on VMware Cluster Technology
by Ying Li
Abstract: In order to overcome the problems existing in the current Internet of things information integration model, Lightweight Information Integration Model of IOT based on VMware Cluster Technology was proposed. This model designs the standard coded identification for each node in the Internet of things, constructs the category of node attributes and node object categories of the Internet of things, and the node LDAP information model, and constructs the node management platform of the Internet of things. VMware cluster technology is introduced, and VMware Infrastructure is used to realize the virtualization and integration of lightweight Internet of things information from server, network and storage. The results show that the number of servers used in this paper is 4, with the least number. The server occupies less space than 8U; The power consumption of the server is lower than 3500W/h, and the power consumption is lower, showing a good performance.
Keywords: VMware cluster technology; Lightweight; IOT information; Integration.
The Research of Human Interaction Recognition Based on Fusion Features of Key Frame Feature Library
by Haiyan Zhang, Shangbing Gao
Abstract: Some issues such as computational complexity and low recognition accuracy still exist in human interaction recognition. In order to solve the problem, the paper has proposed innovative and effective method based on fixed features of key frame feature library. Firstly, GIST feature and HOG feature were extracted from the pre-processed videos. Secondly, training videos with different kinds of actions were clustered by the K-means algorithm respectively to get key frame feature of each action for constructing key frame feature library. And similarity measure was utilized to calculate the frequency of different key frames in every interactive video, and statistical histogram representation of interactive videos were obtained. Finally, the decision level fusion was achieved by using SVM classifier based on histogram intersection kernel to obtain impressive results on UT-interaction dataset. The correct recognition rate of the proposed algorithm is 85%, which indicates that the proposed algorithm is simple and effective than others.
Keywords: GIST Feature; HOG Feature; Key Frame Feature Library; Histogram Intersection Kernel; UT-interaction Dataset.
Research on Stability Modeling of High Speed Electronic Communication Based on Hierarchical Optimal Mining Algorithms
by Lin Tang
Abstract: In order to solve the problems of low accuracy, bad network topological data integrity and low communication transmission efficiency in the current stability modeling methods for high-speed electronic communication, a new stability modeling method for high-speed electronic communication based on hierarchical optimization mining algorithm is proposed. This paper analyzes the transmission characteristics of high-speed signals in the process of electronic communication, constructs a wireless channel model by using hierarchical optimization mining algorithm, and carries out distance correspondence based on the wireless channel model. The stability of high-speed electronic communication is modeled and analyzed by using one-step transition probability matrix, node relative distribution and link duration. Experimental results show that this method has the advantages of high model accuracy, good topological data integrity and high communication transmission efficiency.
Keywords: Hierarchical optimization mining algorithm; High-speed electronic communication; Stability; Modeling and analysis; Transfer probability matrix;.
Self-adaptive Process Optimization Method for SBS Cloud Application based on Reinforcement Learning
by Haiyan Hu, Chang Su
Abstract: To solve the problem of poor comprehensive performance of the traditional SBS cloud application adaptive process optimization method, a new SBS cloud application adaptive process optimization method based on reinforcement learning was proposed. Establish the adaptive action type selection model, realize the optimal choice of operation type, build the cloud application adaptive process optimization model for resource cost, convert the problem into the corresponding mathematical model, solve and make the mathematical model of the algorithm, and obtain the best adaptive process optimization scheme SBS cloud application. The simulation results show that the predicted load value of the method is the closest to the actual load value, the relative error of the prediction is less than 21.03, and the average time is less than 2.3s, indicating that the method has good performance and high practical application value.
Keywords: Enhanced learning; SBS cloud application; self-adaptive process optimization; system performance optimizationrnrn.
Cross-domain Single Sign-on Authentication of Information Security in Network Environment
by Ajun Cui, Wei WANG, Hua-feng ZHANG, Yan-hong MA, Chen LI, Xiao-ming WANG
Abstract: Aiming at the problems of long response time, poor security and information integrity in cross-domain single sign-on authentication research of network security, this paper proposes a cross-domain single sign-on authentication method based on SAML, including the design of login authentication control and revocation authentication scheme. In login authentication, efficient interaction between identity providers and service providers reduces unnecessary steps to achieve single sign-on. At the same time, a revocation scheme based on cumulative function is proposed to ensure that the local root signature is used for system parameter information, thus further ensuring network security. The experimental results show that the login time of the proposed method is below 400ms, and the login response takes a short time. The success rate of the proposed method against false login is over 90%, and the security is good. The data packet loss rate of this method is below 2%, high integrity and reliability.
Keywords: Network security; cross-domain single sign-on; parameter information; revocation of authentication.
Machine Learning Method Based on Improved Drosophila Optimization Algorithm
by Haiying Wang
Abstract: Aiming at the problems of poor classification effect and high CPU ratio of traditional machine learning methods, a machine learning method based on improved drosophila optimization algorithm was proposed.The rank one data mapping and the low order data are established.In low rank support vector set, CP rank organization of traditional support vector machine is used to improve data security.The traditional drosophila algorithm was improved and optimized to increase the number of data iterations, ensure the compatibility of rank one data, improve the optimal calculation of drosophila, and increase the density clustering.The decomposition process is designed to evaluate the objective function value of the optimal solution.In the evaluation process, support vector machine is used to complete the label classification of learning data.Experimental data show that this method performs well in data classification effect, low-rank data storage dimension characteristic performance and CPU operation proportion performance.
Keywords: Machine learning; Drosophila algorithm; Low rank data; Support vector machine.
Research on Blurred Edge Information Segmentation of Image based on Computer Vision
by Xu Zhang
Abstract: In the process of image fuzzy edge information segmentation by traditional methods, the segmentation effect is not ideal, the completion time is long and the accuracy is low.A fuzzy edge information segmentation method based on computer vision is proposed.After image denoising, image sharpening is carried out to extract image fuzzy edge information features.By designing a super-pixel grid, the pixels of the fuzzy edge information features of the image are matched, the inverse tensor information of the fuzzy edge of the image is analyzed, and the multi-threshold values are normalized.The processing results are overlaid on the single object in the image to realize the fuzzy edge information segmentation.Experimental results show that the proposed method has better segmentation effect, shorter completion time and higher accuracy, and is of practical significance.
Keywords: Computer vision; Image blurring; Edge information; Multi-threshold segmentation.
Multi-Tag Cleaning Software For Cyclic Redundant Data Stream Based On Rfid In Cloud Computing
by Yujiang Fei
Abstract: In order to overcome the problems of poor accuracy and recall and low data integrity existing in the current research results of multi-tag cleaning for cyclic redundant data streams, this paper designs a multi-tag cleaning software for cyclic redundant data streams in cloud computing based on RFID. According to the factor graph model, the software models cloud computing network, infers the overall quality of network link connection by cyclic reliability propagation method, and realizes data acquisition by probability graph reasoning. The improved RFID reader reads the information of the tag data to realize the preliminary processing of the cyclic redundant data stream. Multi-label cleaner and RFID reader cooperate to complete multi-label cleaning of cyclic redundant data stream under cloud computing. The experimental results show that the proposed multi-tag cleaning software for cyclic redundant data streams has higher accuracy and recall than the results of references, and has higher data integrity.
Keywords: RFID; Cloud computing; Cyclic redundancy; Data stream; Multi-tag; Cleaning;.
Intelligent Evaluation Model Of E-Commerce Transaction Volume Based On The Combination Of K-Means And Som Algorithms
by Junjie Niu
Abstract: Due to the lack of steps to process sensitive data in the traditional intelligent evaluation model of e-commerce transaction volume, which leads to poor evaluation effect, an intelligent evaluation model of e-commerce transaction volume based on the combination of k-mean and SOM algorithm is proposed.Taking system intelligent clustering as the core, the transaction volume data in the Web log is collected through the collection system.The desensitization rule is used to establish the data transmission model of trading volume.The detection of sensitive data adopts signal detection technology and analytical processing.Based on k-means algorithm, data desensitization is accomplished to achieve data denoising, and a clustering evaluation model is established to complete data mining and analysis.The experimental results show that the model has a good denoising effect, and the ROC curve is closer to the upper left corner with an AUC value of 0.9712, which verifies the effectiveness and superiority of the model.
Keywords: System intelligent clustering; E-commerce transaction volume; Evaluation model; K-means; SOM.
A PSO Based Improved Clustering Algorithm for Lifetime Maximization in Wireless Sensor Networks
by Santar Pal Singh, S.C. Sharma
Abstract: A wireless sensor network (WSN) is comprised of enormous amount of miniature devices known as sensor nodes. Sensor nodes are highly energy restrained due to their battery dependent operations and because of the harsh area deployment it is not feasible to change or revitalize their batteries. For maximizing the network lifetime, energy conservation is a crucial measure of performance improvement of WSNs. So, energy consumption and network lifetime are crucial issues in WSNs. Large portion of energy are consumed in communication process of sensor network. Several routing algorithms have been projected to accurately rout the packet in WSNs. Usually classification of routing algorithms can be done into two categories: based on network structure or based on protocol operations. Based on network structure, these algorithms are broadly divided into flat based, hierarchy based or cluster based, and location based routing algorithms. Due to energy efficiency and network stability, cluster based routing is the best choice. In this paper, we propose a swarm optimization based clustering algorithms for WSNs. We perform the analysis of our new clustering algorithms with LEACH, LEACH-C, and HEED on the basis of performance metrics that contain energy, number of alive nodes, delay, PDR, and throughput. Simulation outcomes confirm that proposed clustering algorithm outperform the other ones.
Keywords: wireless sensor network; clustering algorithm; energy efficiency; network stability; network lifetime.
Virtual Machine Allocation Method For Cloud Computing Based On Multi-Objective Evolutionary Algorithm
by Jinling Ma
Abstract: In order to overcome the problems of data access delay and high energy consumption in virtual machine configuration of cloud computing, a new virtual machine allocation method for cloud computing based on multi-objective evolutionary algorithm is proposed in this paper. Based on multi-objective evolutionary algorithm, the objective of energy consumption and access delay optimization is replaced by an objective model, which is used as the allocation objective model of computational virtual machine. The ant colony algorithm and K-NN local search method are introduced to obtain the optimal solution of the allocation target model and realize the output of the optimal allocation scheme for cloud computing virtual machines. The experimental results show that the access delay and energy consumption of the proposed method are significantly lower than those of other methods, which proves that the proposed method has better performance.
Keywords: Multi-objective evolution; Cloud computing; Virtual machine; Allocation.
Automatic Repair of Pollution Attacks in SDN Topology Network Based on SOM Algorithm
by Xiaoliang Yang, Yi Dai, A. Lun
Abstract: When using the current method to recover the vulnerabilities caused by pollution attacks in SDN network, there are some problems, such as low coverage, low accuracy of vulnerability detection results and low repair efficiency. To this end, an automatic repair method for topology pollution attacks in SDN network based on SOM algorithm is proposed. This paper analyzes the security vulnerabilities in SDN architecture and SDN network, and extracts the attack flow characteristics of topology pollution attacks in SDN network. The extracted attack flow features are input into SOM algorithm to detect and identify vulnerabilities caused by pollution attacks. Combining the theory of vertex concavity and convexity and geometric triangle, the detected vulnerabilities are repaired to realize the automatic repair of topology pollution attacks in SDN network. The experimental results show that the proposed method has high coverage, high accuracy of vulnerability detection and high efficiency of vulnerability repair.
Keywords: SDN network; Pollution attack; Vulnerability detection; Vulnerability repair; SOM algorithm;.
Unlabeled Facial Expression Capture Method In Virtual Reality System Based On Big Data
by Feng Gao
Abstract: In view of the problems of high error rate and low efficiency in the traditional method of facial expression capture without markers, a method of facial expression capture without markers based on large amount of data is proposed. Haar feature is used to determine the initial position of human face, and active shape model is used to extract unmarked facial feature points. The extracted feature points and the generated triangle mesh are tracked by the optical flow tracking method. The displacement of the face feature points is used to promote the overall change of the mesh and complete the unmarked facial expression capture. The experimental results show that the error rate of this method is in the range of 1.2% - 1.7%, the error rate is small, and it needs 20s-34s to capture facial expression, which is more practical and efficient.
Keywords: Big data; Virtual reality system; Unlabeled; Facial expression capture.
BB84 with Several Intercepts and Resend Attacks using Partially Non-orthogonal Basis States through a Depolarizing channel
by Mustapha DEHMANI, El Mehdi Salmani, Hamid Ez-Zahraouy, Abdelilah Benyoussef
Abstract: The paper presents the partially non-orthogonal basis states effect in presence of several intercept and resend attacks through a depolarizing channel. The quantum error and the mutual information are computed for arbitrary angles of the non-orthogonal basis states and the depolarizing channel parameter. It is found that the information security depends strongly on the non-orthogonal basis states angle, the depolarizing channel parameter, the intercepts and resend attack probability and the eavesdropper number. Besides, it is found that for any eavesdroppers number and basis states angle; the protocol is more secure if the depolarization parameter is lower than a threshold value and the maximum value of the quantum error decreases by increasing the depolarizing parameter and / or the eavesdroppers number. While for the depolarization parameter above the threshold value the protocol is not secured independently of any values of the non-orthogonal basis states angle and/or the eavesdroppers number.
Keywords: Quantum cryptography; Intercepts- resend attacks; Non-orthogonal basis states; depolarizing channel; Quantum error; Information Security.
Research On Web Users Behavior Data Mining Based On Feature Orientation
by Hui Zhang, Xiao-ling Jiang, Fa Zhang
Abstract: In view of the problems of long mining time and high error rate in the existing network user behavior data mining methods, a network user behavior data mining method based on feature preference is proposed.The interaction relationship in social network is analyzed as the constraint condition of feature selection.Laplasian operator is used to construct the feature selection model of network user correlation and to quantify the relationship between users.The improved ant colony algorithm is used to obtain the optimal feature subset to realize the network user behavior data mining.The experimental results show that, compared with the traditional methods, the mining time of the proposed method is shorter and the mining error rate is lower under the condition of low and high excavation strength, which verifies the effectiveness of the proposed method.
Keywords: Feature orientation; Web user’s behavior; Data mining.
Research On Global Tourism Information Query Method Based On Association Mining
by Zhongqi Zhao, Yanjie Tan
Abstract: The traditional global tourism information query method can not mine the intrinsic relevance of the discrete information because of its strong discreteness, which results in inaccurate query results and long query response time. To overcome this problem, a global tourism information query method based on association mining is proposed. Using SAS/ME tools, data missing processing, data normalization, correlation analysis and data filtering of global tourism information data in the data are carried out, and the global tourism information data set is constructed according to the results of data processing. This paper uses C4.5 decision tree algorithm and SAS Miner Enterprise to mine the association of global tourism information in data centralization. According to the result of association mining, a global tourism information query system is established to realize the global tourism information query. The experimental results show that the method has the shortest query response time and higher query accuracy, which shows that the method is more suitable for global tourism information query.
Keywords: Association mining; Global; Tourism information; Query; Normalization processing; Discrete attributes;.
Abnormal State Recognition Method for Online Intelligent Examination based on Improved Genetic Algorithm
by Bo Yang, Hongbin Li, Huan Xie, Jianyong Zhao, Rong Zhu, Lei Zhao
Abstract: In order to overcome the defects of the traditional test monitoring scheme, this paper proposes a new method of online intelligent test abnormal state recognition based on improved genetic algorithm.This method sets the test status parameters and defines the criteria of abnormal status, collects the online test information and builds the information database.Image preprocessing is realized from two aspects of image segmentation and grayscale processing.The improved genetic algorithm is used to analyze and collect data intelligently and search image feature points to obtain abnormal feature extraction results.Match the extracted feature results with the established database information to realize the abnormal state recognition results, and start the corresponding abnormal alarm program.The experimental results show that the accuracy of the proposed method is 18.57% higher than that of the traditional method, indicating that the method has a better application prospect.
Keywords: Improved genetic algorithm; Online examination system; Intelligent examination; Abnormal state; State recognition;.
Analysis of Intelligent Test Paper Generation Method for Online Examination based on UML and Particle Swarm Optimization
by Bo Yang, Huan Xie, Kuan Ye, Huan Qin, Rong Zhu, Anchang Liu
Abstract: In order to overcome the problems of unclear objective function and inaccurate optimal solution in traditional test paper generation methods, an online intelligent test paper generation method based on UML and particle swarm optimization is proposed. A mathematical model of test paper generation based on UML modeling tool is established, and the objective function of test paper generation is obtained. The improved particle swarm optimization algorithm is used to solve the objective function of test paper generation. The optimal solution of the objective function is introduced into the question bank. The test questions in the test bank are combined and imported into the online examination system to realize the intelligent test paper formation of online examination. The experimental results show that this method has strong adaptability and can achieve better performance than the traditional intelligent test paper generation method.
Keywords: UML; Particle swarm optimization; Online examination; Intelligent test paper generation; Constrained project indicators; State object matrix.
Research on Dynamic Performance Regulation Method of Bidirectional Relay Network Based on SDN Architecture
by Jianxin Qiu
Abstract: In order to overcome the shortcomings of the traditional dynamic regulation methods of bidirectional relay network, such as low resource utilization and high energy consumption of nodes, a dynamic regulation method of bidirectional relay network performance based on SDN architecture is proposed. SDN architecture is used to construct the bidirectional relay network model. The subcarrier pairing algorithm is used to pair the subcarriers, and then the model of two-way relay network is optimized. The optimal configuration of node power and subcarrier power is carried out respectively, and the joint energy consumption control model of two-way relay network is constructed by using game theory. In this way, the performance of bidirectional relay network can be dynamically adjusted. Through experiments, compared with the traditional method, the proposed method improves resource utilization, reduces node energy consumption, and has better performance.
Keywords: SDN Architecture; Bidirectional; Relay; Performance; Dynamic; Adjustment;.
Intrusion Detection of Hierarchical Distribution Network System based on Machine Computation
by Xiaohong HE
Abstract: In order to solve the problems of low detection accuracy and long detection time of traditional hierarchical distributed system intrusion detection method, a hierarchical distributed system intrusion detection method based on machine computing is proposed. By judging the Chinese protocol type of IP message and the control bit value of TCP, the network traffic is transformed into different sub-flows, and the characteristic parameters of traffic behavior are extracted from the sub-flows. Based on the invasion behavior characteristics obtained, the vector with six - dimensional characteristics is selected as the important symbol of invasion. By combining rule detection method, support vector machine and machine learning classification, the intrusion detection of hierarchical distribution network system is realized by embedding detection modules at different levels. Experimental results show that this method can effectively reduce intrusion detection time and improve detection accuracy.
Keywords: Machine computation; Hierarchical distribution network; System intrusion detection.
Digital English Teaching Resource Sharing System Based On Logical Database
by Fengtian Xu
Abstract: The traditional English teaching resource sharing system has the problems of low distributed computing performance and low data steady-state transmission rate. This paper proposes a design method of digital English teaching resource sharing system based on logical database. The hardware part of the system constructs the logical database of the system, designs the business layer of the logical database in the system, and reads and writes the resources. Establish a database resource search engine, reduce the calculation process and realize the effective use of digital English teaching resources and course recommendation. Using multi type system development software, the system file configuration and service address are developed to complete the system design. The experimental results show that compared with the traditional teaching resource sharing system, the distributed computing performance of the designed system is improved by 37%, and the data steady-state transmission rate is increased by 29%, which has distinct advantages.
Keywords: Logical database; Digitalization; English teaching; Resource sharing; Virtual server.
Research on High Accuracy Mandarin Emotion Recognition Method Based on Local Optimal Algorithms
by Caihua Chen
Abstract: In order to overcome the problem that the traditional Mandarin emotion recognition method does not optimize the result of emotion classification, which leads to low accuracy and recognition efficiency, a high-precision Mandarin emotion recognition method based on local optimization algorithm is proposed. The emotion database of Putonghua is constructed by the dimension emotion model. On this basis, the emotion signals of Putonghua speech are preprocessed and extracted. Through clustering analysis, emotion features are statistically calculated and classified, and local optimization algorithm is used to optimize the classification results to achieve high-precision recognition of Mandarin emotion. The experimental results show that compared with the traditional method, the accuracy and recognition efficiency of the proposed method have been improved to a certain extent, and the shortest recognition time is only 19 ms.
Keywords: Local Optimization Algorithm; High Precision; Mandarin Emotion Recognition; Emotion Feature Extraction.
Construction Of Load Balancing Scheduling Model For Cloud Computing Task Based On Chaotic Ant Colony Algorithm
by Jie Yu
Abstract: In order to overcome the problem of low scheduling balance and long time in traditional load scheduling model for cloud computing task, a load balancing scheduling model for cloud computing task based on chaotic ant colony algorithm is proposed. Task scheduling strategy is selected through task scheduling framework to achieve parallel task scheduling. Based on chaotic ant colony algorithm, cloud computing resources are deployed, and the load objective function of cloud computing task is constructed. Based on the constructed objective function, a load balancing scheduling model for cloud computing tasks is established, thereby achieving load balancing scheduling for cloud computing tasks. The experimental results show that the model has a high scheduling balance, the scheduling time is always less than 5 ms, and the scheduling efficiency is high. This model is more suitable for the balanced scheduling of cloud computing resources, which is feasible.
Keywords: Chaotic ant colony algorithm; Cloud computing; Task load; Balanced scheduling model;.
A Sample De-noising Method for FCM Clustering Induced by Gauss Kernel
by Yunxing Wang, Liyan Li, Zhicheng Wen
Abstract: Aiming at the problem of instability of clustering result because of the existence of noise samples in general FCM algorithm, an improved FCM algorithm induced by Gauss kernel induced is proposed. Firstly, the influence of samples distribution on clustering is analyzed, and the Gauss kernel function is used to nonlinear map of the samples, then the improvement of the objective function of the FCM algorithm and clustering of samples are achieved, thus the purpose of suppressing noise is achieved. The experimental comparison shows that compared with other FCM algorithms, the proposed algorithm's successful classification rate is about 10% higher and the partition coefficient is about 10% lower than other FCM algorithms, indicating that the algorithm has higher clustering effectiveness.
Keywords: Fuzzy C-Means; Gaussian Kernel Function; Noise Sample; Samples.
Research On Personalized Recommendation Method Of Popular Tourist Attractions Routes Based On Machine Learning
by Hui Wang
Abstract: In view of the problems existing in the current tourist route recommendation methods, this paper proposes a personalized tourist attraction route recommendation method based on machine learning.The tourism heterogeneous information data is collected by using Html parser, and the captured data are used as machine learning training samples.The characteristics and user interest characteristics under the rating were extracted, and the target user interest characteristics were taken as the starting point, combining the features of scenic spots and tourist routes.Recommend scenic spots to users to realize personalized recommendation of popular scenic spots.The experimental results show that the method proposed in this paper has a recommendation accuracy of more than 95%, a recommendation time of between 115ms-130ms and a user satisfaction rate of over 89%.With high recommendation accuracy and user satisfaction, and less time to calculate the recommendation, it is a reliable method to recommend tourist routes.
Keywords: Machine Learning; Tourist Attractions Routes; Recommendation.
Study On Geographic Disaster Information Monitoring In Sensitive Areas Based On Spatial Analysis
by Caijian Mo, Fengqiang Wu, Li Chen
Abstract: In order to overcome the problems of low monitoring accuracy and efficiency in the current risk information monitoring methods, this paper proposes a spatial analysis based monitoring method for geographical risk information in sensitive areas. Firstly, this method constructs a real-time monitoring platform with monitoring terminal, information data collection, real-time monitoring and analysis, and designs and analyzes the functional modules. Using spatial analysis technology and target location, the real-time collection of hazard information data and target centroid location are carried out. C / S model is used to manage the geo hazard information data in sensitive areas. The concept of E-R chart is used to build a database, which provides data support for the monitoring process. The experimental results show that the accuracy and efficiency of the proposed method are high, and the highest monitoring accuracy is 98.7%.
Keywords: Spatial analysis; Sensitive area; Geographic disaster information; Monitoring.
Research On Self-Incremental Expansion Method Of Knowledge Base Based On Deep Learning
by QIANG YU, LANLAN LIU
Abstract: In order to overcome the problems of low recall, low precision and long time consuming in current knowledge-base expansion algorithms, a new knowledge-base self-increment expansion algorithm based on deep learning is proposed. The data in the knowledge base is preprocessed by concept stratification, and the running example diagram of the knowledge base is designed according to concept stratification theory. Deep learning tool is used to expand the initial query words of knowledge base, and Word2vec is used to train the document set to build the knowledge base by calculating the cosine similarity. Based on the knowledge base, the noise detection model is constructed by convolution neural network. Through the deep learning, the extended words are filtered to realize the self-expanding of knowledge base. Experimental results show that the proposed algorithm has high recall rate, precision rate, and the algorithm takes less time, which verifies the effectiveness of the proposed algorithm.
Keywords: Deep learning; Knowledge base; Self-incremental expansion; Cosine similarity.
High-Load Reliable Transmission Method For Wireless Network Hybrid Sink Node Data
by Fan-li Meng
Abstract: Data transmission of a hybrid sink node in a wireless network is susceptible to multipath interference, which results in poor output reliability. Therefore, a high-route reliable transmission method of data onto a hybrid sink node in a wireless network is proposed. A channel model is constructed, and a multi-channel cascade matched filtering method is used to suppress data transmission interference. The spatial spectrum feature is extracted and the spectrum beam method is used to control the data transmission load gain. Adopt spatial link equalization configuration to control channel adaptive equalization. The channel spread spectrum model is established to complete the high-route reliable transmission optimization of the data of the wireless convergence node. Simulation results show that this method can provide a higher load for node data, high reliability, and low output error rate, thereby improving the reliability of data transmission.
Keywords: Wireless network; hybrid sink node; data; high load; reliable transmission.
Agimm Tracking Filter Algorithm Based On Maneuvering Feature Correction
by HUANG Qinlong, Xu-fang ZHU, Jia-wei XIA
Abstract: Considering the fast speed and strong maneuvering of small offshore targets, traditional filtering algorithms cannot effectively track the filtering due to the unstable grid center of weak maneuvering target and slow tracking convergence of strong maneuvering target. To overcome this problem, an adaptive grid interaction multiple model (AGIMM) tracking filter algorithm based on maneuvering feature correction was developed. The modified algorithm detects the target maneuvering according to the residual information of the filter, while the model structure can be adjusted under different maneuvering conditions. Monte Carlo simulation shows that the AGIMM algorithm based on maneuvering feature correction can improve the calculation accuracy by 35% and reduce the calculation time by 18%.
Keywords: tracking filter; maneuvering feature; turn model; adaptive grid interaction multiple model; simulation.
Behavior Pattern Recognition of Video Image based on Computer Vision Technology
by Fan Jiang, Yi Yu
Abstract: In order to overcome the problems of low target tracking accuracy, low recognition accuracy and low recognition efficiency existing in traditional behavior pattern recognition methods for video image, this paper proposes a behavior pattern recognition analysis method for video image based on computer vision technology. This method obtains video image information through computer vision technology, and processes the collected video image. Combining with automatic selection tracking region algorithm and particle filter algorithm, the tracking state equation is constructed on the basis of the target color distribution model to track the target in the video image. The feature of video image is extracted, to construct multi-instance learning model. Feature instances are selected, and positive instances are constructed and input into multi-instance SVM classifier to complete the recognition and analysis of video image behavior. The experimental results show that the proposed method takes less than 2 seconds to identify the target, and has high target tracking accuracy, recognition accuracy and recognition efficiency.
Keywords: Computer vision technology; Video image; Human behavior pattern; Recognition method;.
Fault Diagnosis Model Based on Adaptive Generalized Morphological Filtering and LLTSA-ELM
by Jie Xiao, Jingtao Li
Abstract: It is difficult for single feature to contain all the information needed to describe the running state of the equipment, while multi-features can contain more information about running state, but the redundancy between high-dimension features can easily reduce the accuracy of the classifier. Aimed at that,a fault diagnosis method for rolling bearings combining adaptive generalized morphological filter, Linear Local Tangent Space Alignment and Extreme Learning Machine (LLTSA-ELM) is proposed. Firstly, the rolling bearing vibration signals are filtered by adaptive generalized morphological filter. Secondly, the multi-domain features are extracted from filtered signal to construct high-dimensional features set of bearing.Thirdly, the dimension of high-dimensional features is reduced by maximum likelihood estimation (MLE) and LLTSA.Finally, the bearing condition monitoring model based on ELM is constructed by the reduced dimension features and then use it to analyze and diagnose the running state of bearing.
Keywords: Adaptive Generalized Morphological filter; LLTSA; Dimension Reduction; ELM; Fault Diagnosis.
The Color Block Registration Method Of Fuzzy Image Under Multi-Layer P-Spline Geometric Transformation
by Xiaofeng GUO, Jie XU, Wenquan LU
Abstract: In order to overcome the problem that the traditional similarity measure is easy to be affected by the gray-scale migration, this paper proposes a color block registration method of fuzzy image under the multi-layer P-spline geometric transformation. This method mainly sets sparse coding as similarity measure, and divides two images that are not registered into image blocks. K-SVD(K-singular value decomposition) algorithm is used to train image blocks, acquire analysis dictionary and find sparse coefficients. The multi-layer P-spline geometric transformation is used to simulate the fuzzy image, and the gradient descent method is used to optimize the objective function to complete the color block registration of the fuzzy image under the multi-layer P-spline geometric transformation. The experimental results show that the registration time is the lowest, the root mean square error is kept below 0.04, and the registration accuracy is up to 100%.
Keywords: Multi-layer P-spline; Geometric transformation; color block of fuzzy image; Registration method.
Storage Method For Customer Preference Information In E-Commerce Platform Based On Similarity Matching Algorithm
by Chengliang Lin, Lihua Jiang
Abstract: In order to overcome the problems of small IOPs parameters, long response time and inaccuracy of customer access key information similarity in existing e-commerce platform information storage, this paper proposes a new method of customer preference information storage in e-commerce platform. In this method, crawler algorithm is used to collect customer access information, and key customer access information is extracted by setting indicators. Libsvm classifier is used to classify customer access key information, similarity matching algorithm is used to calculate the similarity of customer access key information, and collaborative filtering recommendation algorithm is used to recommend relevant information for customers, so as to realize the storage of customer preference information in e-commerce platform under the background of artificial intelligence. Experiments show that the proposed storage method greatly improves the IOPs parameters, and the minimum storage response time is 21ms, which fully shows that the proposed method has better storage performance.
Keywords: Artificial intelligence; E-commerce platform; Customer preferences; Storage; Information;.
Visual Information Enhancement Method Of Multimedia Human-Computer Interaction Interface Based On Virtual Reality (Vr) Technology
by Li Li
Abstract: The traditional enhancement method of multimedia human-computer interface has the problems of poor user experience and long running time.A visual information enhancement method of multimedia human-computer interface based on virtual reality technology is proposed.The relationship among virtual environment, users and real environment is constructed by using the modeling technology and three-dimensional technology in virtual reality technology to generate multimedia objects; Under the virtual reality technology, the image transformation technology is used to transform the local part of the analog image; The gray level histogram is constructed according to the human visual characteristics. The idea of equalization is to adjust the dynamic range of gray level and enhance the color contrast of image. The simulation results show that the proposed method can solve the problems of blurring, whitening, blocking and ghosting in the interface, and the enhanced visual effect is more in line with the characteristics of human eyes.
Keywords: Virtual reality technology; Multimedia; Human computer interaction; Interface; Visual information; Enhancement.
Research On Fast De Duplication Of Text Backup Information In Library Database Based On Big Data
by Ling Ji
Abstract: In order to overcome the problems of poor effect and low efficiency of traditional information de duplication methods, this paper proposes a fast de duplication method of text backup information in library database based on big data. Firstly, this paper carries out parallel mining of text information features in library database, uses the features with strong classification ability to determine the parameter value of the repeated feature function, obtains the entries with the parameter value higher than the threshold value, determines the number of text repeated backup information and the group weight, sets the difference between the two as the remaining digits, and stops de duplication when the remaining digits are lower than the threshold value. The experimental results show that the average accuracy of this method is 96.95%, the weight removal efficiency is always above 98%, and the weight removal effect is good.
Keywords: Big data; Library; Database; Text Backup Information; De Duplication; Simhash Algorithm.
Research on Traffic Flow Detection Algorithm Based on Road Video
by Jianfeng Han, Yan Zhang, Hongyang Zhang
Abstract: When using the road monitoring video to detect the traffic flow, the vehicle may be treated as one due to the close distance in the video, or the vehicle may be miss detected or missed due to noise interference, resulting in inaccurate traffic data. In order to solve this problem, this paper proposes to generate a foreground for road vehicles by LK local differential optical flow method, calculates the Euclidean distance between the vehicle and the background by using the base watershed algorithm, screens the vehicle position, and segment the adhesion vehicles. At the same time, according to the best position of the vehicle in the video, the virtual detection line is set. The Kalman filter is used to optimize the monitoring data, the traffic flow data statistics error problem is solved, and the accurate traffic flow data is obtained. The experimental result shows that the method mentioned in the paper can obtain accurate traffic data.
Keywords: Vehicle Detection; ,Vehicle Segmentation; Watershed Algorithm; Optical Flow Method.
Framework for concealing medical data in images using modified Hill Cipher, Multi-bit EF and ECDSA
by R. Sreejith, S. Senthil
Abstract: Digitization spread to the core area of the medical field and one of the results is telemedicine. Telemedicine provides the transfer of medical information of patients such as medical results, reports and images which closely deals privacy. The real essence of telemedicine application relays on the secure storage and transfer of medical data over unsecured platform or network where the patient is remotely located from the service provider. The proposed work focused on combining security aspects into a single framework providing the CIA triad of security i.e., Confidentiality, Integrity and Authentication To achieve the above criteria, we use Modified Hill Cipher and multi-bit Encoding Function (EF) for Confidentiality, ROI based Hashing for Integrity and Elliptic Curve Digital Signature Algorithm (ECDSA) for Authentication. The experimental results prove the framework have a better embedding capacity with a comparatively higher PSNR value.
Keywords: Telemedicine; Security; Data Hiding; Steganography; Hill Cipher; Differential Expansion (DE); Multi-bit Embedding; Elliptic Curve; ECDSA.
Optimization of Massive MIMO Data Classification Algorithm Based on Fuzzy C-Means and Differential Evolution Method
by Jia Chen
Abstract: In the environment of massive data multi input multi output (MIMO), it leads to poor data recognizability, clustering ability and overall data processing ability. Therefore, a massive MIMO data classification algorithm based on fuzzy c-means and differential evolution method is proposed. The information flow model of time series of massive data in data flow is established, and the time series features of massive data are extracted by using autocorrelation feature analysis method. Through fuzzy C-means, the features of massive data are cross fused and clustered. The global convergence and stability of data classification are adjusted and controlled by differential evaluation method, and then the delay of data classification is modified to optimize the data classification algorithm. The simulation results show that the data classification recall rate of this method is higher than 95%, and it has the advantages of strong clustering ability and low misclassification rate.
Keywords: MIMO; Massive data classification; Clustering; Information fusion.
3D Scene Reconstruction Method Based on Image Optical Flow Feedback
by Yulin Su
Abstract: When the traditional method is used for 3D reconstruction of a 3D scene, the noise in the scene image cannot be effectively removed, the reconstruction efficiency is low, and the reconstruction accuracy is low. Therefore, a 3D reconstruction method based on optical-flow feedback is proposed to build the camera mathematical model. The weighted least square method is used to denoise the scene image based on the camera mathematical model. Based on the relationship between the optical-flow and the scene flow, the optical-flow is used to calculate the scene flow vector through the grid adjustment algorithm, and the rough grid is adjusted according to the calculation results to realize the 3D reconstruction of the three-dimensional scene. The experimental results show that the proposed 3D scene reconstruction method has good image denoising effect, high reconstruction efficiency and high reconstruction accuracy.
Keywords: Image denoising; Optical-flow feedback; 3D scene; 3D reconstruction.
Enhanced View-independent Representation Method for Skeleton-Based Human Action Recognition
by Yingguo Jiang, Lu Lu, Jun Xu
Abstract: Human action recognition is an important branch of computer vision science. It is a challenging task based on skeletal data because joints have complex spatiotemporal information. In this work, we propose a method for action recognition, which consists of three parts: view-independent representation, combination with cumulative Euclidean distance, and combined model. First, the action sequence becomes view-independent representations independent of the view. Second, these representations are further combined with cumulative Euclidean distances, so the joints more closely associated with the action are emphasized. Then, a combined model is adopted to extract these representation features and classify actions. It consists of two parts, a regular three-layer BLSTM network, and a temporal attention module. Experimental results on two multi-view benchmark datasets Northwestern-UCLA and NTU RGB + D demonstrate the effectiveness of our complete method. Despite its simple architecture and the use of only one type of action feature, it can still significantly improve recognition performance and has strong robustness.
Keywords: Action recognition; enhanced spatiotemporal representation; attention model; Euler angle; cumulative Euclidean distance.
A Deep Learning Based Method for Aluminum Foil Surface Defect Recognition
by Hui WANG
Abstract: In order to effectively detect and classify various defects on the surface of aluminum foil products, including contaminants, coining, shine marks and scratches, etc., the method of convolution neural network?CNN?is used, and the detection of surface defects of aluminum product is realized by machine learning. Firstly, aluminum foil images are collected by CCD camera, and edge detection is performed on these images to obtain a complete picture area. Then, the RPCA (robust principal component analysis) method was used to perform underlying low-rank and sparse decomposition on the data matrix to obtain the defect areas in these images; Finally, using TensorFlow platform to build a CNN network model, loading aluminum foil images and training to get
Keywords: Defect Detection; CNN; Deep Learning; Tensorflow.
Active Contour Model for Image Segmentation Based on Salient Fitting Energy
by Yingyu Ji, Xiaoliang Jiang
Abstract: Although segmentation of image is very important in disease diagnosis, it still exist some difficult problems to precise segmentation, such as noise and intensity inhomogeneity. Aiming at these issues, a novel level set algorithm based on salient fitting energy is presented. We firstly transform original image into a new modality which utilizes gray scale change characteristics of local area. Secondly, a data term of salient fitting energy can be constructed by solving the deviation between new modality and input images in a neighborhood. In addition, distance regularized term is introduced in proposed method to remove re-initialization process. Experiment on a lot of medical and synthetic images demonstrate that the proposed method has good segmentation ability on the part of visual perception.
Keywords: active contour; segmentation; salient fitting energy; level set.
Design Of Power Line Safety Operation And Maintenance Monitoring System Based On Cloud Computing
by Haiying Wang
Abstract: In view of the long response time of current power line operation and maintenance monitoring system, a method of power line safety operation and maintenance monitoring system based on cloud computing is proposed. In the hardware part of the system, the secure access platform is used to connect the mobile terminal and the mobile application server to improve the security of the system. In the software part, the HASH algorithm is used to allocate users of the user interface to realize the security operation and maintenance of the system. The simulation results show that the maximum system response time of the proposed method is 3.21 seconds, while the other methods are higher than 4 seconds. This shows that the proposed method has the shortest time-consuming, improves the efficiency of power line operation and maintenance, reduces unnecessary consumption, and makes China\'s power network management more efficient and reasonable, and meets the strategic needs of the future development of power system.
Keywords: Cloud Computing; Power Line; Security Operation and Maintenance; Monitoring System;.
Research on Adaptive Prediction of Multimedia Data Transmission Path Based on Machine Learning
by Pingping Xiao, Xiaoguang Li, Juan Zhu
Abstract: Aiming at the problems of high probability of data packet loss and long data receiving delay in current methods, an adaptive prediction method of multimedia data transmission path based on machine learning is proposed. This paper analyses the path length of multimedia data transmission, data integrity, energy consumption in the process of data transmission and reception delay, and ranks the advantages and disadvantages of data transmission paths, finally realizes the adaptive prediction of multimedia data transmission paths. The experimental results show that the proposed method is less likely to lose data packets and reduce the data receiving delay when predicting the path of multimedia data transmission. The experimental results verify the effectiveness of the proposed method. At the same time, the accuracy of adaptive prediction of data transmission path is discussed, and the discussion results of high accuracy of path prediction are obtained.
Keywords: Machine Learning; Multimedia; Data Transmission; Path; Adaptive Prediction;.
Research on Precision Playing Method of Website Advertising Based on Machine Learning
by Zhuomin Huang
Abstract: In order to solve the problems of poor accuracy and low degree of freedom in website advertising, a precise method of website advertising based on machine learning is proposed. The output vector of the previous hidden layer is added to the output vector of the input layer as the input vector of the hidden layer; the output vector of the hidden layer is obtained by using the excitation function for the input vector of the hidden layer at a certain time; the error calculation of the above-mentioned results and the ideal output results is carried out based on the error. The experimental results show that the method has high precision of placement, and the degree of freedom varies between 224 and 248, which indicates that the method has low complexity of advertising placement.
Keywords: Machine Learning; Advertising; Precision; Delivery.
Design of E-Commerce Cluster Information Classification and Extraction System Based on Relevant Vector Machine
by Yanjun Sun
Abstract: Aiming at the problem of low accuracy and long time-consuming in traditional classification and extraction system, this paper studies and designs an electronic commerce clustering information classification and extraction system based on correlation vector machine. In order to enhance the expansibility of the whole system, the hardware part of the system adopts modular design. The core modules are: information preprocessing module, information representation module and classifier module. The classification extraction algorithm based on Relevant Vector Machine (RVM) is selected as the core algorithm of the whole system. In order to improve the extraction speed of clustering information classification in e-commerce and improve the algorithm, a calculation method of excluding the worst category in each round of comparison is proposed. On the basis of protecting the classification accuracy of the original algorithm, the extraction time of information classification is effectively reduced. The simulation results show that the classification and extraction results of the designed system are aggregated in the range of -0.4 to 0.9, and the classification and extraction accuracy is high. The extraction time changes in the range of 0.1s to 0.2s, which can quickly and accurately complete the classification and extraction of e-commerce clustering information.
Keywords: Relevance Vector Machine; E-commerce Clustering; Information Classification and Extraction; System Design\r\n.
The Information Security Transmission Method For Intelligent Examination Based On Zigbee Communication
by Huan Xie, Bo Yang, Zhigang Ren, Kebin Mu, Xueqian Zhao, Bangyan Li
Abstract: In order to overcome the problems of low network utilization, poor network transmission security and high key maintenance cost existing in the traditional information security transmission method for intelligent examination, the information security transmission method for intelligent examination based on ZigBee communication is proposed. The network node obtains the main secret key through pre installation, and realizes the network key and connection key by using asymmetric key system, that is, using public key and private key. Public key encryption and private key decryption are used to verify the trust center. The sending node uses the trust center to get the public key of the target node, and the target node uses the private key to decrypt the received information, so as to achieve the purpose of information security transmission of intelligent examination. The simulation results show that the proposed method effectively improves the network utilization rate, and also enhances the comprehensive performance of information security transmission of intelligent examination.
Keywords: ZigBee communication; Intelligent examination; Information security; Transmission method.
Research On User Experience Evaluation Of Man-Machine Interaction Interface Based On Virtual Reality Technology
by Hong SONG, Zhi YUE
Abstract: Due to the complexity of user experience evaluation information data of human-computer interaction interface, the traditional evaluation method has a single information channel, resulting in the low quality of human-computer interaction interface. This paper proposes a user experience evaluation method based on virtual reality technology.Using virtual reality technology to restore the quality of human-computer interaction interface, build the fuzzy analytic hierarchy process of samples, and obtain the corresponding membership function. According to the superposition of the membership degree, the membership degree evaluation matrix of the first level index is established, and the user experience evaluation of human-computer interaction interface based on virtual reality technology is completed. The simulation results show that the proposed method can effectively improve the operation efficiency and accuracy, and has good stability, and has good application value.
Keywords: Virtual reality technology; Man-machine interaction interface; User experience; Evaluation.
An Identification Method Of Malicious Nodes In Wireless Communication Based On Dynamic Reputation Algorithm
by Jia Chen
Abstract: Due to the complex internal structure of wireless communication network, the traditional methods for malicious node identification are relatively single, which leads to a large number of security risks in the network environment. This paper proposes a method of identifying malicious nodes in wireless communication based on dynamic reputation algorithm. A model of WSN wireless communication malicious node identification based on routing protocol reputation mechanism is established. The network is divided into clusters to determine the transmission path of network packets. Send the packet to the sink node and analyze it, calculate the node number and reputation value in the packet and compare with the threshold value to realize the identification of malicious nodes in wireless communication. The simulation results show that the proposed method can complete the identification of malicious nodes in wireless communication with high accuracy, and it takes less time and has better recognition performance.
Keywords: Dynamic reputation algorithm; malicious nodes in wireless communication; Identification.
Load Balancing Selection Method For Heterogeneous Power Communication Equipment Based On Software Defined Network
by Yingjie Jiang, Chao Ma, Qiusheng Yu, Youxiang Zhu, Lei Liu
Abstract: In order to overcome the problem of poor load balance of power heterogeneous communication equipment, this paper proposes a SDN (Software Defined Network) based load balance selection method for power heterogeneous communication equipment, which divides the architecture into three levels: application, control and infrastructure. This paper introduces the load balance structure of distributed and centralized integration, and uses the centralized control of wireless access point to realize the unified control of communication equipment terminal access. Taking the minimum mean square error of load rate and the minimum total number of resources occupied as the objective function, the genetic algorithm is introduced to solve the optimal solution of the objective function, real business and network connection relevance, so as to obtain the load of power heterogeneous communication equipment. The experimental results show that the total time-consuming is about 9s, and the average blocking rate is 0.04.
Keywords: SDN; Power; Heterogeneous; Communication equipment; Load balancing.
A Recognition Method For Visual Image Of Sports Video Based On Fuzzy Clustering Algorithm
by Rongai Sun
Abstract: In order to overcome the problem of poor recognition accuracy of action visual image, a method of sports video action visual image recognition based on fuzzy clustering algorithm is proposed. This method uses the fuzzy clustering algorithm to segment the action visual image of sports video, which is divided into foreground and background. In the image foreground, the edge contour of the object is cut through the edge detection step, and the action representation feature quantity extraction method based on the joint point is used to extract the action feature of the human joint point in the foreground, so as to complete the action visual image recognition of sports video. The experimental results show that the segmentation accuracy is high, the noise iteration performance is good, and the recognition accuracy is higher than 0.96, which can achieve high-precision recognition of sports video action image.
Keywords: Fuzzy clustering algorithm; Sports video; Visual image; Recognition.
3D Reconstruction Of UAV Remote Sensing Sequence Image Based On Iterative Constraint Weighting
by Tiebo Sun, Meng Li, Weibing Wang, Chunyue Liu
Abstract: Aiming at the problem that the traditional 3D reconstruction method of UAV remote sensing sequence images takes time and affects the reconstruction accuracy, a 3D reconstruction method of UAV remote sensing sequence images based on iterative constraint weighting is proposed. Construct a UAV remote sensing platform, and process the images of UAV remote sensing sequences through image enhancement, uniform light processing and stitching. An iterative constraint weighting method is introduced to solve the global rotation matrix problem as a rotation vector in algebra. Through the iterative constraint weighting method, the second programming obtains the optimal solution of the global position and optimizes the global position and attitude. According to the position and attitude parameters of the acquired UAV remote sensing sequence image and the reconstruction point cloud, the 3D reconstruction of the image is realized. Experimental results show that the method is short, accurate, effective and reliable.
Keywords: Iterative constraint weighting; Remote sensing sequence image; 3D reconstruction.
R2DCLT: retrieving relevant documents using cosine similarity and LDA in text mining
by R.S. Ramya, Ganesh Singh, Santosh Nimbhorkar Sejal, K.R. Venugopal, S.S. Iyengar, L.M. Patnaik
Abstract: The availability of digital documents over web has increased exponentially and hence there is a need for effective methods to retrieve and organise. Since data is dispersed globally and unorganised, a number of algorithms have been proposed based on relevance calculations. However, it is found that there is a gap between users search intention and retrieved results. In this paper, we propose a framework for retrieving relevant documents using cosine similarity (CS) and LDA in text mining (R2DCLT) The uniqueness of this approach is that LDA is applied for the documents and extracted patterns like unigram, bigram and trigram. Documents are ranked based on the CS score. Experiments are conducted on Reuters Corpus volume and custom news dataset. It is observed that R2DCLT outperforms pattern taxonomy and relevance feature discovery models by providing high quality relevant documents with improved response time and dynamically updated document set.
Keywords: pattern mining; query search; query expansion; text feature extraction; text mining.
Community Structure Detection Algorithm Based on Link Prediction
by Baomin Xu
Abstract: Community structure identification has received a great effort among computer scientists who are focusing on the properties of complex networks. The label propagation algorithm is a near linear time algorithm to find a good community structure. Despite various subsequent advances, an important issue of this algorithm is the efficiency and accuracy of the identified community structure. In this paper, we propose a novel community detection algorithm by using link prediction algorithm based on label propagation. The method is the first to introduce the idea of link prediction into community detection. The experimental results show that the proposed method is less resolution limited than modularity optimizing methods, and it can be more effective in detecting communities.
Keywords: Complex networks; Link prediction; Label propagation; Community detection.
A Novel Protocol of RFID Tag Identification Using a Single Mobile Reader
by Xiaowu Li, Runxin Li, Lianyin Jia, Jiaman Ding, Jinguo You, Yingying Lü
Abstract: Classic RFID tag anti-collision protocols mainly aim at the identification of tags in small area. That is, all unidentified tabs are located in the reader's read-write region. If there is one or more tag outside the reader coverage area, multi-reader RFID systems is usually used to accomplish the tag identification task. However, the cost of multi-reader RFID systems is much greater than that of a single reader mode. There are many other tag identification scenes where tag distribution area is only slightly larger than a reader coverage area. For the scene, in the paper, we let all tags, including outside the reader coverage area, have an opportunity to enter the reader coverage area and to be identified by moving the reader near them. The scene is called single mobile reader systems at the paper. Single mobile reader systems cause the tag reappearance phenomenon (TRP) and the multiple identification problem of tag (MIPT). TRP and MIPT decrease the effective tag identification efficiency seriously if using existing tag identification protocol. In the paper, we propose an improved EPC C1G2 protocol, which can increase effective tag identification efficiency effectively.
Keywords: RFID; Anti-collision; Tag identification; Single Mobile RFID Systems; Effective Tag Identification Efficiency (ETIE).
Achievement Management System For University Students Based On Cloud Storage Technology
by Mei Xu, Yi Liu
Abstract: In order to overcome the problems of long storage time, long query response time and low data aggregation accuracy of current methods for achievement data management of university student, this paper proposes an achievement data management system of university student based on cloud storage technology. Cloud storage technology is used to construct cloud storage system architecture, and design the overall architecture and software logical structure of the achievement management system. Combined with functional requirements, functional modules are built. E-R model is used to design database, to achieve the design of university students achievement data management system. The experimental results show that the data storage time is less than 0.5s, the query response time is less than 0.3us, and the accuracy of data aggregation is higher than 80%. It proves that the management system can meet the users experience requirements.
Keywords: Cloud storage technology; University students; Achievement data management system;.
Target Similarity Matching Algorithm Of Big Data In Remote Sensing Image Based On Henon Mapping
by Qing Sun, Quanyuan Wu
Abstract: In order to overcome the problem of low matching accuracy in traditional big data target similarity matching algorithm of remote sensing image, this paper proposes a new target similarity matching algorithm based on Henon mapping. The randomness of big data target in remote sensing image is analyzed by using the variation of Henon mapping invariant distribution. According to the randomness, the track of big data target in remote sensing image is selected to build a two-layer similarity matching model. The first layer of the model uses coarse granularity to reduce the dimension of big data, and the second layer uses the fine-grained representation of similar track set to output several tracks similar to the big data target of remote sensing image, so as to achieve the target similarity matching. The experimental results show that the proposed method has high matching accuracy, and the highest matching accuracy can reach 99.7%.
Keywords: Henon mapping; Remote sensing image; Big data target; Similarity matching.
Computer Threat Information Filtering Algorithm Based On Fusion Difference
by Liang Qian
Abstract: In order to overcome the problems of low information filtering coverage and long time-consuming in traditional computer threat information filtering methods, a new computer threat information filtering algorithm based on fusion difference is proposed. In this algorithm, the user feature model is established by integrating the difference method, and the objective function is introduced to protect the difference privacy data in the computer. Through Chinese word segmentation, feature extraction and weight calculation, user requirements and threshold initialization, filter matching, user feedback, five steps to achieve computer threat information filtering. The experimental results show that compared with the traditional algorithm, the proposed algorithm has a higher recall rate and a shorter filtering time, the shortest filtering time is only 0.4min, which shows that the proposed algorithm has a higher practical value.
Keywords: Fusion difference; Computer; Threat information; Filtering algorithm;rnrn.
The Self-Increasing Expansion Method For Knowledge Space Based On Deep Learning Algorithm
by Yuanhan Weng, Jingan Wang
Abstract: In order to overcome the problems of the traditional expansion method for knowledge space, such as small expansion range and low accuracy, this paper proposes an expansion method for knowledge space based on deep learning algorithm. Through deep learning algorithm, combined with multi-modal information fusion method, including the fusion and expansion of the current knowledge space, the knowledge space expansion framework is constructed. The framework is set as space organization knowledge, knowledge indexing, knowledge navigation, knowledge retrieval and other parts, and knowledge division is realized according to the continuous classification of knowledge sequence information. In the extended space, the multi-structure state of knowledge element is integrated by semantic description technology to realize the expansion of knowledge space. Experimental results show that the expansion method for knowledge space based on deep learning algorithm is better.
Keywords: Deep learning; Knowledge expansion; Self-increasing;.
Dynamic Fault Diagnosis means of the Power message System Based On Big Data
by Dong He, Tong Chen, Haichao Huang, Weihao Qiu, Yize Tang, Jinxia Jiang
Abstract: Aiming at the poor fault diagnosis ability of traditional power information system, a dynamic fault diagnosis method based on big data for power information system is proposed. Firstly, the original fault information of power information system is sampled, and the collected fault characteristic data are reconstructed by multi feature and information fitting. Then, the attribute distribution detection and big data mining are carried out for the fault dynamic characteristics of power information system. According to the high-order spectrum feature distribution of the extracted power information system fault signals, the dynamic fault diagnosis and fuzzy clustering analysis are carried out for the power information system, and the fault diagnosis is optimized according to the classification results. The simulation results show that the dynamic fault diagnosis accuracy of power information system is high, the fault sample detection results are accurate and reliable, and the dynamic fault detection ability is improved.
Keywords: Big data; electricity message system; fault; dynamic diagnosis; detection.
Automatic anti-interference control of intelligent mechanical communication terminal based on Neural Network
by Xiao-Xing Shi, Yanqin Zhang, Li-Ye Liu
Abstract: In order to overcome the disturbance of mechanical intelligent communication terminal caused by mechanical continuous movement, which affects the communication effect and quality, an active disturbance rejection control method for mechanical intelligent communication terminal based on neural network is proposed. This method analyzes the working principle of Cmnn and combines it with active disturbance rejection control technology. Aiming at the initial disturbance, the active disturbance rejection control technology is the dominant control, and the Cmnn enters the learning state and optimizes the weights according to the period; the active disturbance rejection control technology participates in the disturbance control, eliminates the interference, and ensures that the mechanical intelligent communication terminal is in a stable state. The experimental results show that the steady-state control time is about 0.06s, and the tracking error is controlled in the range of [+ 5 ?rad, - 5?rad], which has higher tracking accuracy and robustness.
Keywords: Cerebellar model; Neural network; Intelligent mechanical communication terminal; ADRC; Response speed.
Construction Of Integration Model For Regional Sub-Meter High-Resolution Remote Sensing Geographic Information
by Caijian Mo, Fengqiang Wu, Li Chen
Abstract: In order to overcome the problem of poor data coordination existing in the existing geographic information integration model, a new regional sub-meter high score remote sensing geographic information integration model construction method is proposed. The application of distributed integration technology to build a regional sub-meter level high score remote sensing geographic information integration model. Joint processing of images, in the integrated environment of GIS and RS, transform the independent variables in the data into a vector grid integrated structure, establish a joint index format based on B+ tree index and quadtree index, complete the RS and GIS model Build. Experimental results show that the model can apply street, city name, and place type keywords, with better integrated indexing effect and higher data coordination.
Keywords: Remote sensing geographic information; Integrated model; Grid; Joint index;.
Personalized Recommendation Method Of English Micro-Lectures Teaching Resources Based On Internet Of Things Platform
by Zhengui Zhang
Abstract: In order to overcome the problems of low accuracy and long time-consuming in traditional teaching resource recommendation methods, a personalized recommendation method of English micro course teaching resources based on Internet of things platform is proposed. Collect the Internet of things platform server agent, server and client for user interest resource data, preprocess user interest resource data. The user interest model is constructed by using the obtained user interest resource data, and the English micro class teaching resource model is constructed by vector space model. This paper combines the user interest model with the English micro class teaching resource model, and makes personalized recommendation of English micro class teaching resources. The experimental results show that: the accuracy rate of the proposed method is as high as 98%, and the recommendation time is less than 6 s, which can realize the personalized recommendation of English micro class teaching resources.
Keywords: Internet of things platform; English micro-lectures; Teaching resources; Personalized recommendation; Vector space model; Recommendation algorithm.
Feature Extraction Method Of Multi-Frame Image In Cloud Video
by Siyuan Cheng, Shouren Diao, Shuo Cai, Hongyan Liu
Abstract: In order to solve the problem of low detection and recognition ability of multi frame images in cloud video, a feature extraction method of multi frame images in cloud video is proposed in this paper. The image region information distribution structure model of multi frame images in cloud video is established, and the image edge gradient information detection method is used to detect the features of multi frame images and pixel information fusion in cloud video. According to the region information of the image, the edge contour feature detection model of multi frame image in cloud video is established, and the edge gradient information of multi frame video image is segmented and reconstructed by active contour detection method. Simulation results show that the method has high accuracy and high degree of information fusion, and improves the ability of video image detection and adaptive recognition.
Keywords: cloud video; multi-frame image; feature extraction; detection and recognition.
Feature Space Fusion Classification Of Remote Sensing Image Based On Ant Colony Optimization Algorithm
by Qing Sun, Quanyuan Wu
Abstract: In order to overcome the problems of low classification accuracy and poor application effect of traditional remote sensing image feature space fusion classification method, a new remote sensing image feature space fusion classification method based on ant colony optimization algorithm is proposed. According to the ant colony algorithm state transition rule, the global optimal path is updated. The spatial structure, edge and texture features of remote sensing image are extracted by feature extractor. The fusion weight coefficient of remote sensing image space and spectral feature vector is calculated. The extracted remote sensing image feature vector is replaced by the maximum likelihood method Image classification discriminant formula is used to realize remote sensing image feature space fusion classification. The experimental results show that the average classification accuracy is improved by 9.75%, and the classification speed is improved by 15.6%, which effectively improves the image recognition rate.
Keywords: Ant colony optimization algorithm; Remote sensing image; Image feature; Feature space fusion; Image classification;.
Design of real-time monitoring system for transmission channel energy consumption in wireless sensor networks
by Yong XIAO, Jingfeng YANG, Yun ZHAO, Xin JIN
Abstract: The collection of energy consumption data of the transmission channel of the wireless sensor network is neglected, which has the problem that the operation rate of the network is low and the time is long. The new real-time monitoring system of transmission channel energy consumption for wireless sensor network is proposed and designed. For the hardware part, the structure of the network master node, the child node and the transit node are analyzed respectively, employing the single chip program to design hardware part of the system;In order to complete the design of the software part of the system, a network transmission channel is designed to query historical data, display data in real time and collect energy consumption data in real time.The experimental results show that the proposed system has a higher network transmission rate and shorter monitoring time, which verifies the real-time and effectiveness of the system.
Keywords: wireless sensor network; transmission channel; energy consumption; real-time monitoring system;.
Fastvgg Network And Its Application In Automatic Identification Of Traffic Signs
by Xinyuan Li, Shangbing Gao, Zhonghe Lu, Kaige Gui
Abstract: Traffic sign recognition is an important part of driver assistance systems. Because the types of traffic signs are complex and diverse, they are difficult to identify. Traditional recognition methods require manual extraction of features, which is difficult and inaccurate. This paper proposes a FastVGG network based on VGG neural network to extract the features of the target image to realize the recognition of traffic signs under different angles and illumination. In the method of this paper, the connection layer parameters and the number of network layers are moderately reduced, the merging step is increased, and the recognition speed is improved. When the parameter value is less than zero, the Leaky ReLU is used to replace the activation function to solve the problem of neuron death. The experimental results of the German Traffic Sign Recognition Data Set (GTSRB) show that the algorithm can achieve accurate classification while increasing the speed.
Keywords: Traffic sign recognition; driving assistant system; deep learning; VGG Net.
Research on the Enhancement of Internet UI Interface Elements based on visual communication
by Can Zhou
Abstract: Aiming at the poor vision of traditional UI interface, this paper proposes a new method to enhance the communication of visual elements of multi-icon UI interface in mobile Internet. Design multi-icon user interface through user research, feature analysis and environment analysis. Principal component analysis (PCA) was used to obtain feature vectors and collect visual elements of mobile Internet multi-icon UI. The multi-resolution and multi-scale Retinex algorithm is applied to the communication of color enhancement visual elements in the multi-icon UI interface. Experimental results show that the mobile Internet multi-icon UI designed by the method in this paper has bright colors, and the average time for users to watch the mobile Internet multi-icon UI enhanced by communication of visual elements reaches 7.6 seconds, which verifies the effectiveness of the method in this paper.
Keywords: Mobile; Internet; Multi-icon; UI interface; Visual elements; Communication enhancement.
An Image Recognition Method For Speed Limit Plate Based On Deep Learning Algorithm
by Jian Gao
Abstract: In order to overcome the problems of large number of sample data collected from speed limit image and unclear image feature hierarchy, an recognition method for speed limit plate image based on deep learning algorithm is proposed. This method combines deep learning with SVM to build a multi-level classification model, and uses deep learning method to re represent the original data through unsupervised learning. The image features of speed limit plate are extracted in depth, and the image is preprocessed by color component compensation, image denoising and threshold segmentation. The similarity between the image features extracted layer by layer and the standard features is calculated, and the final recognition result of speed limit plate image is obtained through feature matching. The experimental results show that the average recognition rate is increased by 6.6%, which can effectively provide data reference for vehicle speed control in the actual driving process.
Keywords: deep learning algorithm; speed limit plate; image recognition; SVM;rnrn.
Research on the Selection of Interconnection Modes of Asymmetric Internet Backbone
by Qiming Tang, Meijuan Li, Qianbing Xiao
Abstract: At present, the interconnection mode between internet backbones in China is only peering, which makes the small and medium-sized backbone networks at a disadvantage in the interconnection. Optimizing the interconnection mode between Internet backbone in China has become an urgent problem to be solved by our government regulatory agencies. By constructing a dynamic game model, this paper analyzes the optimal selection of the interconnection modes of the asymmetric Internet backbone networks. The analysis results show that the selection of interconnection modes between the asymmetric backbone is related to the difference in network scale between the backbone networks. When the difference between the scale of the weak backbone network and the dominant backbone network is small, the paid peering mode could be chosen; when the scale of the weak backbone network and the dominant backbone network is large, the transit mode should be chosen. The introduction of transit mode has promoted competition between backbone networks. According to the conclusion of the study, specific suggestions are put forward to optimize the interconnection modes between internet backbone in order to promote the healthy and rapid development of the internet industry in China under the background of the triple play and national cyber development strategy.
Keywords: Internet Backbone; Interconnection; Paid Peering; Transit.
Research on multi-channel data acquisition system of production index information based on genetic algorithm
by Yilun Zhang
Abstract: The multi-channel data acquisition system is the key link of the Industrial Computer Tomography (ICT) in image processing. How to realize the real-time data acquisition and storage of these huge data content has become an urgent problem for ICT. Based on this, a multi-channel data acquisition system of production index information based on genetic algorithm is proposed. This paper introduces the design of the basic structure of the multi-channel data acquisition system of production index information, expounds the hardware design and software design of the multi-channel data acquisition system, and gives the design and implementation scheme of the system. It is confirmed by experiments that the multi-channel data acquisition system based on genetic algorithm designed in this paper has high acquisition accuracy. It is particularly excellent, especially in the same sampling rate, compared with the traditional multi-channel data acquisition system of production index information. Through further analysis, it is found that the trigger mode of the multi-channel data acquisition system based on genetic algorithm can significantly reduce the loss rate of hardware resources compared with the traditional multi-channel data acquisition system.
Keywords: genetic algorithm; production index information; multi-channel; data acquisition.
Optimization of computer virtual image reconstruction method based on feature point matching
by Chiyu Pan
Abstract: Traditional computer virtual image reconstruction methods have the problems of slow reconstruction speed and insufficient image clarity. To solve this problem, an optimization method of computer virtual image reconstruction based on feature point matching is proposed. SURF operator is used to extract image feature points. After obtaining feature points, image feature points are matched according to TZNCC constraints. The virtual image is reconstructed by sparse method, and the high resolution depth image in virtual vision is represented by dictionary sparse linear combination. The sparse coefficient of the image is solved by alternating direction multiplier algorithm, and the problem of virtual image reconstruction is transformed into a problem of solving sparse signal, so that a better reconstruction effect can be obtained. The experimental results show that the proposed method has high speed and clarity of image reconstruction.
Keywords: Feature point matching; Computer; Virtual image; Reconstruction.
Image Segmentation Using Active Contour Model Driven by RSF and Difference of Gaussian Energy
by Qile Zhang, Xiaoliang Jiang
Abstract: As we all know, the region scalable fitting method is sensitive to initializations and suffers from bad results in images with complex scene. In our article, we put forward a new framework by integrating region scalable fitting (RSF) term and difference of Gaussian (DOG) term for segmenting images. We first propose a DOG function which can enhance the contrast at the edges of objects. Then, the RSF energy term is introduced to drive the curve closer to the edge. In the next step, the regularization term is established which can avoid of the process of re-initialization. Compared with traditional classical methods, the proposed technique is more flexibility with initialization and has better segmentation performance.
Keywords: Difference of Gaussian; Region scalable fitting; Active contour; Image segmentation.
Automatic assessment of adherence of terms and conditions of web service based process
by Lingaraj Panigrahy, Laxmi Narayan Padhy, Ajaya Kumar Tripathy
Abstract: In general, web service based processes (WSBP) are distributed and composed of one or more WSBP (called component web services (CWS)) which are developed and controlled by third parties. Hence, the functional and quality guarantee terms made with the WSBP provider and client are depends on the correctness of all CWSs involved. To enhance the trust and reliability on WSBP, it is essential to make functional and quality terms compliance assessment of adherence at run-time. Moreover, functional and quality need and priority constraint may change over a period of time. A run-time, dynamic and nonintrusive approach for compliance assessment can handle this situation. This article proposes a functional and quality guarantee terms specification mechanism between theWSBPprovider and client. The conformance verification of execution traces of WSBP against the guarantee terms specification is enabled by translating these specifications to automata.Acompiler is designed to automate the process of translation of these automatons into executableCprograms. These auto-generated C programs monitors the WSBP and CWS interactions to detect violation of functional guarantee terms or validate of quality related guarantee terms at runtime ofWSBP. This mechanism of conformance verificationworks independent of the WSBP design and implementation process. The use of the proposed approach is demonstrated by applying in an application system.
Keywords: Web Services; Web Service Composition; Service Monitoring; Formal Language.
A Multi-Attribute Recognition Method Of Vehicles Line-Pressing In Parking Lot Based On Multi-Task Convolution Neural Network
by Shaohui Zhong, Ting Hu
Abstract: In order to solve the problems of low recognition accuracy and long recognition time, a multi-attribute recognition method based on multi task convolution neural network is proposed. The structure principle of multitask convolution neural network is analyzed, and multitask is set in the bottom area of convolutional neural network. The Hough transform is used to extract the parking line in the parking lot, and the input layer of the multi-attribute label structure is established by multi-attribute classification convolution neural network. The loss function of vehicle line pressing attributes in different parking lots is obtained by combining the full connection layer and the connecting sub layer. The multi-attribute recognition of vehicle pressure line is realized by measuring and learning the line voltage attributes of vehicles. The experimental results show that the method can effectively identify the line pressing situation of vehicles in parking lot, and the recognition accuracy can reach 99%.
Keywords: Multi-task; convolution neural network; parking lot; line-pressing of vehicle; multi-attribute; recognition method.
Research on online-offline information resource joint regulation method of cross-border e-commerce model based on genetic algorithm
by Yajie Zhao
Abstract: In order to overcome the problems of large adjustment error, low utilization rate of information resources and low execution efficiency caused by traditional methods without considering task execution priority, this paper proposes a joint adjustment method of online and offline information resources based on genetic algorithm in cross-border e-commerce mode.Combined with grid technology, this method sets up an online-offline joint regulation model of information resources through multi-objective planning, and then USES multi-objective selection method to form information subgroups, and obtains new genetic population through crossover and mutation calculation, so as to obtain the optimal scheme of information resources joint regulation. The experimental results show that the maximum implementation efficiency of this method can reach 98.1%, the utilization rate of information resources is always above 95%, and the adjustment error is always below 6%, which proves that this method is effective.
Keywords: Cross border e-commerce mode; online-offline information; information resources; joint regulation; genetic algorithm.
Study On The Method Of Identifying Diseases With Abnormally High Signals Based On Machine Learning Technology
by Xiang SUN, Qianmu LI
Abstract: In order to improve the recognition level of abnormal high signal disease, a recognition method of abnormal high signal disease based on machine learning technology is proposed. The abnormal high signal feature extraction method based on machine learning is used to obtain the abnormal high signal. The wavelet threshold method is introduced to remove the noise signal and extract the features. The method based on BP neural network is used to identify disease types. The results show that the identification performance of this method is obviously better than that of similar methods, and the complexity of identification process is only 1.23%. The anti-interference ability of abnormal high signal in the identification of benign and malignant diseases is as high as 0.99, which can effectively eliminate the interference of noise signal, extract the characteristics of abnormal high signal, and complete the disease recognition.
Keywords: Machine learning technology; abnormally high signal; identification method; characteristic extraction; neural network.
Research On The Effect Of Mobile Multimedia Advertising Based On Deep Learning
by Bin Liu
Abstract: In order to solve the problems of inaccurate analysis and unsatisfactory evaluation effect of traditional methods, this paper proposes a mobile multimedia advertising delivery effect research based on deep learning. Taking the real data set of mobile multimedia advertising as the research object, the feature data of mobile multimedia advertising are extracted by convolution neural network, and the mobile multimedia advertising feature data are classified, and the classification decision model is constructed. The extracted feature data are standardized preprocessed, and principal component analysis is introduced to remove the redundancy of feature data; the evaluation index system of mobile multimedia advertising effect is constructed; and the support vector regression model is constructed to evaluate the effect of mobile multimedia advertising. The simulation results show that the proposed method can accurately evaluate the effect of mobile multimedia advertising, and the evaluation accuracy rate can reach 98%, and the evaluation time is short.
Keywords: Deep learning; Mobile multimedia; Delivery effect; Convolutional neural network; Principal component analysis method; Evaluation index.
Designing a cost-effective model leveraging serverless computing to provide weather forecasts to farmers in rural India.
by Ravi Prakash Varshney, Dilip Kumar Sharma
Abstract: Serverless computing has been widely adopted as a platform for the deployment of various event-driven applications and services. The features like autoscaling, no resource management, pay-as-you-go, less expensive are some of the key reasons for its wide adoption in software development. It allows the developers to define applications as a sequence of event-triggered functions. Weather forecast plays a vital role in Indian agriculture, but rural Indian farmers are devoid of any form of weather forecast information. The dearth of weather information accounts for the reduction in crop productivity and hence, financial loss. In this paper, we propose a cost-effective and scalable architecture and a prototype of a serverless application that provides the weather forecast to the farmers in the form of a mobile push notification or Short Message Service (SMS). With an operating cost of around $25 per month, the proposed model caters effectively to around 100,000 users'. We demonstrate how serverless computing, along with the other services offered by Amazon Web Services (AWS), can be glued to build a scalable and cost-effective solution and assist farmers in mitigating the losses incurred.
Keywords: Serverless computing; Function as a Service (FaaS); Weather forecast; Cloud Computing; AWS Lambda; Amazon web services (AWS).
Exact Outage Performance of Two NOMA Users in Small-cell network over Nakagami-m fading under imperfect CSI
by Chi-Bao Le, Tu-Trinh Nguyen, Thuan Do
Abstract: We consider a two-user system in scenario of relay cooperative non-orthogonal multiple access (NOMA) network with imperfections of channel state information (CSI). Firstly, on the same allocated power source, the base station (BS) transmits the superposition signal to near user and far user, which is analyzed via Nakagami-m fading channels. In this situation, the near user with good channel conditions employs NOMA scheme to serve the far user. Secondly, the far user combines signals from direct link and relay link. Then, the outage performance is examined in closed-form expressions to exhibit performance gap among two NOMA users.
Keywords: non-orthogonal multiple access (NOMA); Nakagami-m fading; outage probability
Inter-contact delay and location information-based routing with adaptive threshold buffer management for delay tolerant networks
by Savita, D.K. Lobiyal
Abstract: Most of the routing algorithms in delay tolerant networks use multi-copy approach and messages stay for long time in the buffer because of no end-to-end connectivity between source to destination node. On one hand these activities ensure the high delivery probability, while on the other hand they introduce high resource overheads per message in the network. Therefore, limited storage capacity of buffer bound us to use intelligent message control buffer management policies. In this paper, we have proposed a combination of buffer management scheme for location information and inter-contact-based routing approach. The proposed buffer management scheme is a combination of message scheduling and message drop policy grounded on the concept of inter-contact delay based on delivery probability, hop count and number of copies of a message in the network. We also used adaptive threshold scheme which depends on average transferred bytes in the network to prioritise messages. It has been observed that our schemes results in reduction of overheads as compared to previous schemes. The probabilistic analysis of location dependent and independent encounters of nodes has been made to gain better insights of encounter timings. Experimental results obtained demonstrate the effectiveness of our proposed scheme.
Keywords: delay tolerant networks; DTNs; power control; opportunistic networks; overhead; buffer management.
Research on automatic annotation of pathological image detail information based on machine learning
by Xiang Sun, Qianmu Li
Abstract: Aiming at the inaccuracy of pathological image diagnosis results of traditional models, the machine learning technology is introduced and a new pathological image diagnosis method is then proposed. In this method, the accuracy of traditional pathological image diagnosis is transformed into the accuracy of detailed information annotation of pathological image. Pathological image pooling is completed through convolutional neural network, and some pathological images with dimensional blurring phenomenon are sharpened. On the basis of the pooling and sharpening of pathological image, the features of pathological image based on convolutional neural network are extracted. And then machine learning technology is used to annotate the extracted pathological image features and complete the automatic annotation of pathological image details. The experimental results show that the proposed model can improve the accuracy of pathological image feature annotation, and can accurately detect the pathological image of cancer cells, providing a new solution for pathological image diagnosis.
Keywords: pathological image; machine learning; information annotation; automatic detection.
Deep features-based dialect and mood recognition using Assamese telephonic speech
by Mridusmita Sharma, Kandarpa Kumar Sarma
Abstract: Learning aided methods are popular for designing automatic speech recognition (ASR) systems. Majority of works have used shallow models in combination with mel frequency cepstral coefficients (MFCC) and other features for speech recognition applications. Although these shallow models are effective but incorporating deep features in the mechanism for speech processing applications is necessary to increase the efficiency. Despite of considerable amount of works on the design of deep learning topologies and training paradigms in supervised domain, very few works have concentrated on deep features which are essential to capture detailed information of speech. This work focuses on the generation of deep features using stacked auto-encoder for normal and time shifted telephonic speech samples in Assamese language with mood and dialect variations. Experimental results show that the deep features learned by the stacked auto-encoder performs better while it is configured for Assamese speech recognition with mood and dialect variations.
Keywords: deep features; auto-encoder; Assamese language; dialects; moods; recognition; artificial neural network; ANN.
MISO assisted multiple access by removing orthogonal: enabling D2D transmission and performance analysis
by Minh-Sang Van Nguyen, Thi-Anh Hoang, Dinh-Thuan Do
Abstract: As a breakthrough system model adopted in the fifth generation (5G) wireless networks, multiple input single output (MISO)-based non-orthogonal multiple access (NOMA) has been introduced and investigated. With device-to-device transmission taken into account, we consider a NOMA-based downlink network under Rayleigh fading in this paper. To look main metric, we study the system outage behaviour, and close-form expressions for the exact outage probability of two NOMA users are obtained. By further evaluating the outage probability under different power allocation factors, fairness in NOMA still is guaranteed. Finally, numerical results are presented to demonstrate the validity of our analysis and show the advantages of NOMA scheme.
Keywords: multiple input single output; MISO; non-orthogonal multiple access; NOMA; transmit antenna selection; TAS.
Multi-agent approach for data mining-based bagging ensembles to improve the decision process for big data
by Ahmed Ghenabzia, Okba Kazar, Abdelhak Merizig, Zaoui Sayah, Merouane Zoubeidi
Abstract: Today, data growth is accelerating to create a big data in various fields, such as social media, websites, e-mails, finance, and medicine. It needs analysis and knowledge extraction. In addition, data mining is a technology whose purpose is to promote information and knowledge extraction from a big data. In this paper, a multi-layered approach based on agents is proposed to extract knowledge from big dataset with bagging algorithm. To achieve this, we call the paradigm of a multi-agent system in Hadoop to distribute the complexity and processing of large datasets across several autonomous entities called agents. The goal is to predict the target class or value for each case in the data using the bagging technique that is dedicated to the task of classification or regression. This proposition will help decision-makers to take right decisions and provide a perfect response time by the use of the multi-agent system in Hadoop. Therefore, to implement the proposed architecture, it is more convenient to use the Apache Hadoop framework, Apache Spark MLlib framework for building scalable machine learning algorithms and JADE platform which provides a complete set of services and agents.
Keywords: big data; Apache Hadoop; Apache Spark MLlib; multi-agent system; MAS; bagging; JADE.
An effective secure data retrieval approach using trust evaluation: HBSEE-CBC
by Rosy Swami, Prodipto Das
Abstract: In recent times, secret data are outsourced into the cloud to a great extent. But, for the security and privacy of the data, it is essential to encrypt the data. Since search using encrypted query is very difficult, hence, facilitating an encrypted data searching scheme is supreme prominence. Though many efficient encryption algorithms are utilised, still it has some issues like increased encryption, decryption and key-generation time. An effective secure data retrieval approach (HSBEE-CBC) is introduced in this work, to avoid these issues. Here, the data are encrypted using elliptic curve cryptography (ECC) algorithm. The encrypted data are clustered using cosine based clustering (CBC) and transferred to the server. For improving the security, trust is evaluated for the user who intends to retrieve the information. After evaluating the trust, the data is decrypted in order to retrieve it. The performance of the HSBEE-CBC is analysed and compared with traditional approaches.
Keywords: cloud computing; information retrieval; elliptic curve cryptography; ECC; cosine similarity; trust evaluation.