Skip to main content
main-content
Top

About this book

This two-volume set (CCIS 1137 and CCIS 1138) constitutes the proceedings of the Third International Conference on Cyberspace Data and Intelligence, Cyber DI 2019, and the International Conference on Cyber-Living, Cyber-Syndrome, and Cyber-Health, CyberLife 2019, held under the umbrella of the 2019 Cyberspace Congress, held in Beijing, China, in December 2019.

The 64 full papers presented together with 18 short papers were carefully reviewed and selected from 160 submissions. The papers are grouped in the following topics: cyber data, information and knowledge; cyber and cyber-enabled intelligence; communication and computing; cyber philosophy, cyberlogic and cyber science; and cyber health and smart healthcare.

Table of Contents

Frontmatter

Communication and Computing

Frontmatter

A Markov Approximation Algorithm for Computation Offloading and Resource Scheduling in Mobile Edge Computing

Mobile edge computing has become a key technology in IoT and 5G networks, which provides cloud-computing services in the edge of the mobile access network to realize the flexible use of computing and storage resources. While most existing research focuses on network optimization in small-scale scenes, this paper jointly considers the resources scheduling of servers, channels and powers for mobile users to minimize the system energy consumption. It’s an NP-hard problem which can only be solved through the exhaustive search with complexity of exponential level. The lightweight distributed algorithm proposed in this paper based on the Markov approximation framework can make the system converge to an approximate optimal solution with only linear level complexity. The simulation results show that the proposed algorithm is able to generate near-optimal solutions and outperform other benchmark algorithms.

Haowei Chen, Mengran Liu, Yunpeng Wang, Weiwei Fang, Yi Ding

A Geographic Routing Protocol Based on Trunk Line in VANETs

To make full use of historical information and realtime information, a Trunk Road Based Geographic Routing Protocol in Urban VANETs (TRGR) is proposed in this paper. This protocol aims to solve the problem of data acquisition in traditional trunk coordinated control system. Considering the actual physical characteristics of trunk lines, it makes full use of the traffic flow of the trunk lines and the surrounding road network, provides a real-time data transmission routing scheme, and gives a vehicle network routing protocol under this specific condition. At the same time, the TRGR protocol takes into account the data congestion problem caused by the large traffic flow of the main road, which leads to the corresponding increase of the information the flow of the section, and the link partition problem caused by the insufficient traffic flow. It introduces different criteria for judgment and selection, which makes the TRGR protocol more suitable for the application of coordinated control of the main road in the urban environment. Simulation results show that the TRGR protocol has better performance in end-to-end delay, delivery rate and routing cost under the scenario of urban traffic trunk lines comparing with other IOT routing protocols. TRGR protocol can effectively avoid data congestion and local optimum problems, effectively increase the delivery rate of data packets, and is suitable for routing requirements in this application scenario.

Di Wu, Huan Li, Xiang Li, Jianlong Zhang

Sub-array Based Antenna Selection Scheme for Massive MIMO in 5G

With rapidly increased throughput demand, operators are rapidly improving coverage and capacity with cost effective techniques in wireless communication network. Developments in technology enables advanced antenna system to be scalable across 5G and future wireless networks. Massive MIMO based advance antenna selection techniques provide powerful and affordable methods that are effective approaches for coverage and capacity of consumers. Prerequisites of an optimal communication system grow quickly, and therefore operators require more facilities to meet their needs. It is necessity to serve many operators and various devices at the same time in the integrated zone, while providing fast speed and consistent performance, makes it the boosting technology yard to meet the requirements of the 5G era. In this paper, we propose a Sub-array based Antenna Selection Scheme (SASS) for massive MIMO based on sub-array switching architecture which is beneficially helpful to achieve optimal throughput, energy efficiency and capacity. Moreover, SASS is cost effective technique which reduces the overall cost of system including computational, communication, and hardware impairments. We have validated our work using MATLAB and results are compared for spectral efficiency when number of antennas and Signal to noise ratio are varied. Results prove the dominance of SASS over counterparts.

Hassan Azeem, Liping Du, Ata Ullah, Muhammad Arif Mughal, Muhammad Muzamil Aslam, Muhammad Ikram

A Green SWIPT Enhanced Cell-Free Massive MIMO System for IoT Networks

This paper investigates the downlink performance of a green simultaneous wireless information and power transfer (SWIPT) enhanced cell-free massive multiple-input multiple-output (MIMO) system where the IoT devices are served by the virtual massive MIMO constituted by a large amount of distributed single antenna access points (APs). On the premise of that all the IoT devices can decode the information and harvest energy from the received signals, the closed-form expressions of downlink spectral efficiency (SE) and harvested energy of this system are firstly derived under the time switching (TS) and power splitting (PS) scheme. After that, to maximize the lowest SE of the IoT devices, a joint optimization problem which takes into account the power control coefficients and energy harvesting coefficients simultaneously is formulated and an optimal bisection based algorithm is proposed to solve it. Besides, the network load management problem is also studied to reduce the fronthaul burden and an AP selection method which considers the imbalance of AP transmission ratio is proposed. Simulation results show that compared with the benchmark methods, our proposed schemes can guarantee higher transmission rates for IoT devices as well as a reduction of load for fronthual links.

Meng Wang, Haixia Zhang, Leiyu Wang, Guannan Dong

Non-orthogonal Multiple Access in Coordinated LEO Satellite Networks

Non-orthogonal multiple access (NOMA) has been widely considered to improve the spectral efficiency in terrestrial wireless networks. In this paper, we extend the idea to satellite networks and propose a NOMA-based scheme for coordinated low Earth orbit satellite systems, where the beam-edge user is supported by two satellites. By exploiting the difference of the equivalent downlink channel gains, users located both at the beam-center and beam-edge can be served simultaneously using NOMA. It is shown that the NOMA-based cooperative method is capable of providing a higher system capacity while guarantee the rate quality of the beam-edge user.

Tian Li, Xuekun Hao, Guoyan Li, Hui Li, Xinwei Yue

Multi-sensor Data Fusion Based on Weighted Credibility Interval

The Dempster-Shafter combination rule often get wrong results when dealing with severely conflicting information. The existing typical improvement methods are mostly based on the similarity of attributes such as evidence distance, similarity and information entropy attribute as evidence weight correction evidence itself. Ultimately, the final weights of the evidences are applied to adjust the bodies of the evidences before using the Dempster’s combination rule. The fusion results of these typical methods are not ideal for some complex conflict evidence. In this paper, we propose a new improved method of conflict evidence based on weighted credibility interval. The proposed method considers the credibility degree and the uncertainty measure of the evidences which respectively based on the Sum of Absolute Difference among the propositions and the credibility interval lengths. Then the original evidence is modified with the final weight before using the Dempster’s combination rule. The numerical fusion example has verified that the proposed method is feasible and improved, in which the basic probability assignment (BPAs) to identify the correct target is 99.21%.

Jihua Ye, Shengjun Xue, Aiwen Jiang

A Locality Sensitive Hashing Based Collaborative Service Offloading Method in Cloud-Edge Computing

Benefiting by the big data produced by ever increasing IoT devices, big data services are gaining popular attention in many areas. However, general IoT terminals are unable to execute these services due to the exponentially growing data and the limited computing resources. And a possible solution is to execute the services on remote cloud data centers. However, transferring all data to remote cloud for process brings huge energy consumption and congestion on the backends under high load conditions. The development of edge servers makes it possible to handle some simple tasks on edge servers. Towards this end, it is imperative to design a collaborative service offloading scheme to process data of complex big data services on both edge servers and clouds. In this paper, to protect user’s privacy and quickly decide offloading destination for big data services, we propose a locality sensitive hashing based allocating strategy called Loyal. Loyal relies on E2LSH technique to hash and encrypt the sensitive data information. In addition, Loyal is able to retrieve suitable service that can be offloaded to the ES in a short time. Finally, the performance of Loyal is presented by simulation experiment.

Wenmin Lin, Xiaolong Xu, Qihe Huang, Fei Dai, Lianyong Qi, Weimin Li

A Testbed for Service Testing: A Cloud Computing Based Approach

Although simulation is an important tool in studying new emerged networks, tests on a real system are still the most ultimate way to validate the capability of a newly emerged network. Traditional tests have their drawback, because of the coupling of test service and test devices. The arise of new network architecture, industry 4.0 and the internet of things, also brings new challenges. Therefore, a testbed for service testing and data process based on cloud computing is proposed. In this testbed, a test is realized as a service hosted by one or several cloud hosts, i.e., test as a service (TaaS). The relationship of different kinds of services in a testbed is modeled, based on the Lotka-Volterra model. The testbed implementation is demonstrated, based on cloud computing, using a typical video test as an example. It is proved that our testbed for service testing and data process is practicable and effective.

Qinglong Dai, Jin Qian, Jianwu Li, Jun Zhao, Weiping Wang, Xiaoxiao Liu

Improve the Efficiency of Maintenance for Optical Communication Network: The Multi-factor Risk Analysis via Edge Computing

Optical communication networks carry people’s communication services. As the structure of communication networks becomes more complex, and the volume of services is growing. It is greater to ensure that the network is functioning properly. Therefore, people are deeply concerned about the maintenance of communication networks. The nodes in the network are an important component of the network. Therefore, it is necessary to conduct the risk analysis on nodes and perform intelligent maintenance based on the results. The normal operation of nodes in a communication network is influenced by many aspects. In order to access the risk of nodes, the risk analysis algorithm proposed in this paper considers network topology, service type, maintenance, environmental factors and the load of the node. Then intelligent maintenance is performed based on the results of the node’s risk analysis, and the edge computing server is introduced to perform the risk analysis on the node in time. So as to improve the timeliness of updating information. According to the results of simulation, maintenance personnel can get more accurate information about the nodes or aspects that need to be maintained, and because of the use of edge computing structures, the method has become more efficient.

Yucheng Ma, Yanbin Jiao, Yongqing Liu, Hao Qin, Lanlan Rui, Siqi Yang

Edge Computing for Intelligent Transportation System: A Review

To meet the demands of vehicular applications, edge computing as a promising paradigm where cloud computing services are extended to the edge of networks can enable ITS applications. In this paper, we first briefly introduced the edge computing. Then we reviewed recent advancements in edge computing based intelligent transportation systems. Finally, we presented the challenges and the future research direction. Our study provides insights for this novel promising paradigm, as well as research topics about edge computing in intelligent transportation system.

Qian Li, Pan Chen, Rui Wang

Consensus Performance of Traffic Management System for Cognitive Radio Network: An Agent Control Approach

The Spectrum sharing is an important topic in Cognitive Radio Sensor Networks (CRSNs). Bio-inspired consensus-based schemes can provide lightweight and efficient solutions to ensure spectrum sharing fairness in CRSNs. This paper studies the consensus performance of traffic management system in Cognitive Radio Networks (CRNs). Research focused on significant topics of spectrum management in scalable CR ad hoc networks such as mobility, sharing, and allocation driven by local control. First of all, CRN is analyzed and is considered as a network of multiagent systems with a directed graph. Secondly, this multiagent problem is transformed into the multi-input and multi-output model of cognitive radio users in the frequency domain. Thirdly, the consensus condition of CR users is proposed based on their information sharing in the network. Moreover, the communication delay is also considered and a delay margin criterion is derived to guarantee the performance of information sharing for CR users. From simulation results, the use of proposed schemes shows low complexity and effectiveness for spectrum sharing processes in randomly deployed CRSNs also shows the effectiveness of the proposed method.

Muhammad Muzamil Aslam, Liping Du, Zahoor Ahmed, Hassan Azeem, Muhammad Ikram

CyberLife 2019: Cyber Philosophy, Cyberlogic and Cyber Science

Frontmatter

An Efficient Concurrent System Networking Protocol Specification and Verification Using Linear Temporal Logic

In critical computer-based systems, safety and reliability are of principal concern, especially when dealing with concurrent transactions on which mobile systems depend on, such as the emerging Internet of Things (IoT). We present a protocol to ensure safety and reliability of systems where concurrent modification of data on routers in a network is possible, by detecting cycles in the conflict graph and ensuring the system is free of any cycle in an effective manner. The existence of a cycle in a conflict graph means that the schedule of such concurrent transactions cannot be serialized. We use temporal logic in the representation of this protocol model to ensure the safety of systems. Administrative routing protocols benefit significantly from this protocol model.

Ra’ed Bani Abdelrahman, Hussain Al-Aqrabi, Richard Hill

Performance Evaluation of Multiparty Authentication in 5G IIoT Environments

With the rapid development of various emerging technologies such as the Industrial Internet of Things (IIoT), there is a need to secure communications between such devices. Communication system delays are one of the factors that adversely affect the performance of an authentication system. 5G networks enable greater data throughput and lower latency, which presents new opportunities for the secure authentication of business transactions between IIoT devices. We evaluate an approach to developing a flexible and secure model for authenticating IIoT components in dynamic 5G environments.

Hussain Al-Aqrabi, Phil Lane, Richard Hill

PAM: An Efficient Hybrid Dimension Reduction Algorithm for High-Dimensional Bayesian Network

In recent years, machine learning has been gradually widely applied to the big data in medical field, such as prediction and prevention of disorders. Bayesian Network has been playing an important role in machine learning and has been widely applied to the medical diagnosis field for its advantages in reasoning under uncertainty. But as the rise of the number of interest variables, the Bayesian Network structure search space is growing super-exponentially. Aiming at improving the efficiency of finding the optimize structure from the large search space of high-dimensional network, in this paper we propose a method, PAM, which is applied to Bayesian Network learning to constrain the search space of high-dimensional network. Several Experiments are performed in order to confirm our hypothesis.

Huiran Yan, Rui Wang

Indoor Activity Recognition by Using Recurrent Neural Networks

Because of the development of the ageing population, most countries are facing an increasingly serious pension resources problem. With the development of Internet of Things, the integration of smart home and smart retirement provides a new solution for the new smart home for the elderly, to achieve the elderly to intelligently support the elderly. This paper is based on the development of this background, mainly to solve the problem of indoor activity recognition of the elderly, so as to prepare for the construction of smart medical care. The specific research process is to process the sensor data collected from the smart environment, identify different activities using RNN, LSTM and GRU models with strong ability to process time series data, realize the target of activity recognition.

Yu Zhao, Qingjuan Li, Fadi Farha, Tao Zhu, Liming Chen, Huansheng Ning

Petal-Image Based Flower Classification via GLCM and RBF-SVM

Flower identification is a difficult problem in practice. Because there are over 250,000 different kinds of species worldwide so far. Even an experienced flower expert needs reference book to categorize a flower because of the high intra-class variation and inter-class similarity. In this study, an automatic flower recognition method was proposed based on digital image processing and artificial intelligence for petal image. Gray level co-occurrence matrix was employed as the image feature and a support vector machine was trained as the classifier. Three different kernel functions were tested and radial basis function performed best. Experimental results revealed that our approach can achieve state-of-the-art classification performance.

Zhihai Lu, Siyuan Lu

A Convolutional Neural Network-Based Semantic Clustering Method for ALS Point Clouds

Point clouds semantic clustering is an important technique for extracting information from point clouds. A large number of point clouds datasets now can be obtained using airborne laser scanning (ALS). Existing clustering methods of point clouds are mostly focus on regular objects and restrictions, these may lead to errors when the scenario is complex or they are segmentation not semantic clustering. In recent years, CNN based methods for point clouds data have achieved good results. So this paper proposed an new method for ALS point clouds semantic clustering using convolutional neural network (CNN). In the approach, firstly the feature of every point with its adjacent points are extracted and transformed into corresponding pixel, the whole point cloud are transformed into single image. Then the image is used as input of a model that based on CNN and superpixel segmentation to achieve semantic clustering goal. We evaluate our method on the public datasets provided by the 2019 IEEE GRSS Data Fusion Contest and compared with common methods for point cloud clustering. The method performs good, shows the potential of deep-learning-based methods in semantic clustering of point clouds.

Zezhou Li, Tianran Tan, Yizhe Yuan, Changqing Yin

Aligning Point Clouds with an Effective Local Feature Descriptor

Point cloud registration is a crucial step and gaining more importance in many challenging 3D computer vision tasks including 3D reconstruction, autonomous navigation, 3D object recognition and remote sensing. In this work, we proposed a highly discriminative local feature descriptor named Local Point Feature Histogram (LPFH) for 3D point cloud registration. LPFH formulates a simple and comprehensive histogram for surface representation, which encompassed a 3D descriptor. Based on the proposed LPFH, we use Random Sample Consensus (RANSAC) algorithm to form our coarse registration stage, followed by an Iterative Closest Point (ICP) fine registration stage, these two steps form our registration algorithm. Validations and comparisons with other point cloud registration algorithms showed that LPFH is low-dimension, efficient, effective and easy to compute.

Xialing Feng, Tianran Tan, Yizhe Yuan, Changqing Yin

A Tutorial and Survey on Fault Knowledge Graph

Knowledge Graph (KG) is a graph-based data structure that can display the relationship between a large number of semi-structured and unstructured data, and can efficiently and intelligently search for information that users need. KG has been widely used for many fields including finance, medical care, biological, education, journalism, smart search and other industries. With the increase in the application of Knowledge Graphs (KGs) in the field of failure, such as mechanical engineering, trains, power grids, equipment failures, etc. However, the summary of the system of fault KGs is relatively small. Therefore, this article provides a comprehensive tutorial and survey about the recent advances toward the construction of fault KG. Specifically, it will provide an overview of the fault KG and summarize the key techniques for building a KG to guide the construction of the KG in the fault domain. What’s more, it introduces some of the open source tools that can be used to build a KG process, enabling researchers and practitioners to quickly get started in this field. In addition, the article discusses the application of fault KG and the difficulties and challenges in constructing fault KG. Finally, the article looks forward to the future development of KG.

XiuQing Wang, ShunKun Yang

An Attention-Based User Profiling Model by Leveraging Multi-modal Social Media Contents

With the popularization of social media, inferring user profiles from the user-generated content has aroused wide attention for its applications in marketing, advertising, recruiting, etc. Most existing works focus on using data from single modality (such as texts and profile photos) and fail to notice that the combination of multi-modal data can supplement with each other and can therefore improve the prediction accuracy. In this paper, we propose AMUP model, namely the Attention-based Multi-modal User Profiling model, which uses different tailored neural networks to extract and fuse semantic information from three modalities, i.e., texts, avatar, and relation network. We propose a dual attention mechanism. The word-level attention network selects informative words from the noisy and prolix texts and the modality-level attention network addresses the problem of imbalanced contribution among different modalities. Experimental results on more than 1.5K users’ real-world data extracted from a popular Q&A social platform show that our proposed model outperforms the single-modality methods and achieves better accuracy when compared with existing approaches that utilize multi-modal data.

Zhimin Li, Bin Guo, Yueqi Sun, Zhu Wang, Liang Wang, Zhiwen Yu

Face Anti-spoofing Algorithm Based on Depth Feature Fusion

With the development of face recognition system towards automation and unsupervised, illegal intruders have become a serious threat to face authentication system by disguising face authentication system, how to ensure the security of the face recognition system has become an urgent problem in face recognition technology. Therefore, living face detection has become an important issue that must be solved in the face authentication system. By deeply studying the importance of facial image color feature information for human face detection, a deep feature fusion network structure is constructed by deep convolutional neural networks ResNet and SENet to effectively train the involved face anti-spoof data. The feature with large amount of information, while suppressing the features with low usefulness, the experimental results are greatly improved compared with the traditional methods, and have higher recognition effect and accuracy.

Jingying Sun, Zhiguo Shi

Facial Micro-expression Recognition Using Enhanced Temporal Feature-Wise Model

Automatic facial micro-expression recognition is challenging for the subtlety and transience in facial motion, and limited databases. Most researches focus on handcrafted techniques for facial micro-expression analysis on two-dimensional images. However, spatiotemporal facial feature representation is a critical issue for facial micro-expression recognition due to its short duration and subtle facial movement. To deeply extract the appearance characteristics and facial changes effectively from facial image sequences, a feature-wise deep learning model was proposed by applying temporal Convolutional Neural Network (3D-CNN) and Long Short-Term Memory (LSTM) to enhance temporal feature learning. There are two stages involved: (1) The CNN was extended to convolute along spatio and temporal simultaneously, to better represent the facial texture and motion. (2) The feature vector obtained by 3D-CNN was fed into LSTM for temporal enrichment. It was demonstrated that the proposed model achieved promising good performance on CASME II and SMIC databases on person-independent and cross-database experiments.

Ruicong Zhi, Mengyi Liu, Hairui Xu, Ming Wan

Dynamic Facial Feature Learning by Deep Evolutionary Neural Networks

Facial Action Coding System is a comprehensive and anatomical system which could encode various facial movements by the combination of basic AUs (Action Units), and makes the emotion categories much wider. Recently, deep learning has been shown its superiority on recognition tasks. Despite the powerful feature learning ability of deep learning, there are still several problems remained. Firstly, a large amount of training data is needed to fully extract features and avoid overfitting. Secondly, the parameters optimization of deep neural network is complex, and the direct guidance of the results is insufficient. In this paper, a spatiotemporal self-learning method is designed by evolutional deep neural network model, and spatial augmentation is utilized to deal with the two problems facing in practical application. The proposed method is conducted on AUs analysis task which is important for emotion identification. The 3D convolutional neural network which could learn dynamic facial features from AUs image sequences is optimized automatically for the topology and hyper-parameters by evolutional scheme. Extensive experiments demonstrated the effectiveness of EVONET (Deep Evolutionary Neural Networks) on the facial databases over alternative methods, including 3DCNNs (3D Convolutional Neural Networks), and several convolutional neural network based models.

Ruicong Zhi, Caixia Zhou, Tingting Li

Baseball Pitch Type Recognition Based on Broadcast Videos

In this paper, we report our work on baseball pitch type recognition based on broadcast videos using two-stream inflated 3D convolutional neural network (I3D). To improve the state-of-the-art of research, we developed our own high-quality dataset, trained and tuned the I3D model extensively, primarily combating the problem of overfitting while still trying to improve final validation accuracy. In the end, we are able to achieve an accuracy of 53.43% ± 3.04% when oversampling and 57.10% ± 2.99% when not oversampling, which is a significant improvement over the published best result of an accuracy of 36.4% on the same six pitch type classes.

Reed Chen, Dylan Siegler, Michael Fasko, Shunkun Yang, Xiong Luo, Wenbing Zhao

Semi-automated Development of a Dataset for Baseball Pitch Type Recognition

In this paper, we report our work on developing a new dataset for baseball pitch type recognition based on youtube videos of the US Major League Baseball games. The core innovation is a largely automated procedure to extract relevant clips from the full game, and automatically label the clips by aligning the infographic information included in the broadcast and the PitchF/X data. We adopted the Needleman-Wunsch algorithm to address the challenges imposed by the aligning the two streams of data based on pitch speed, i.e., minimize gaps and mismatches between the two streams. Manual inspection is used only to select games that include infographic information for clip extraction and to remove erroneous clips for improve the quality of the dataset.

Dylan Siegler, Reed Chen, Michael Fasko, Shunkun Yang, Xiong Luo, Wenbing Zhao

Ford Vehicle Classification Based on Extreme Learning Machine Optimized by Bat Algorithm

The application of automobile identification in life is more and more extensive, so research on related technologies is receiving widespread attention. This article focuses on research on Ford vehicle identification, the theoretical method of identification is proposed and its effectiveness is verified in experiments. We first obtain the side-view image of the Ford car. Secondly, we use gray level co-occurrence to extract the feature of Ford car. Third, we use extreme learning machine as the classifier. Finally, we use bat algorithm to optimize the algorithm, and employ 10-fold cross-validation to ensure the validity of the data. The results of the research indicate that in the same kind of research, the method we employ has the highest accuracy (84.92 ± 0.64%).

Yile Zhao, Zhihai Lu

Design and Implementation of a Wearable System for Information Monitoring

In view of the limited measurement accuracy and short data transmission distance of existing smart wearable devices, this paper designs a wearable multi-information monitoring system. It achieves multi-directional observation of human body via energy consumption, vital signs and location-tracking information. The overall system consists of three parts: data terminal, LoRa communication network and remote monitoring center. Considering the constraint of power and memory cost, the system adopts a set of appropriate chips for hardware design. As the core of the system, data terminal contains a main control module, a signal acquisition module, and a display and transmission module. The microcontroller controls each sensor to collect different original signals which are performed noise reduction, and then these signals are processed by the algorithm embedded on chips to create the final information for monitoring human body. Test results show that the wearable multi-information monitoring system meets the demand for the measurement accuracy of human body information.

Qi Zhao, Tongtong Zhai

A Review of Internet of Things Major Education in China

With the continuous expansion of the Internet of things industry, more and more problems are exposed, among which professionals shortage is one of the most important problems. In order to solve the shortage of IoT professionals, the Chinese government encourages colleges and universities to open IoT major. Due to the short opening time, there are still some problems in professionals training. This paper reviews the situation of Internet of things major education in China. First, we introduce the background of development of Internet of things major, and then through the analysis of the layer models, we put forward a universally used IoT Major’s knowledge structure and related courses list. Finally, this paper classifies the universities offering Internet of things major in China and analyzes their educational status from the perspective of education level, geographical location and curriculum system.

Yuke Chai, Wei Huangfu, Huansheng Ning, Dongmei Zhao

CyberLife 2019: Cyber Health and Smart Healthcare

Frontmatter

Research on the Influence of Internet Use on Adolescents’ Self

The paper explored the influence of Internet use on adolescents’ self-construal and body esteem. Experiment 1 investigated the relationship between Internet use behavior and self-constructional types of 886 adolescents by questionnaire and Self-constructional Scale. The results show that the interdependent-self have a stronger motivation to use the Internet to communicate in order to keep in touch with others and integrate into the group. And they pay more attention to their privacy. However, the independent-self tend to update their status on the Internet, who don’t care too much about the self-image management on the Internet or whether the published status leaking personal information. But none of the result reached significant levels. Experiment 2 investigated the influence of Internet use on body esteem of 240 middle school students by questionnaire and Body Esteem Scale. The results show that there is a significant difference in body esteem on genders. The correlation between boys’ Internet use and body esteem is significant and positive, especially in the dimension of physical attractiveness and upper body strength. There is no significant correlation between Internet use and body esteem in girls.

Huimei Cao, Jiansheng Li

Breast Cancer Risk Assessment Model Based on sl-SDAE

In recent years, the incidence of breast cancer among women in China has increased year by year and it has become the most common malignant tumor in women in China. There are already breast cancer risk assessment models for women in Europe and the United States. However, there is no effective breast cancer risk assessment model suitable for women in China. The paper established an effective breast cancer risk assessment model. It selected the survey data of breast cancer population in China as a data set. The paper combines SDAE and LSTM to build a model based on deep learning methods. It uses the roc curve as an indicator of the experimental results. Experiments show that the model has better performance than traditional machine learning algorithms.

Xueni Li, Zhiguo Shi

Prediction Model of Scoliosis Progression Bases on Deep Learning

By deep learning technique, we present a new approach to model idiopathic single curve scoliosis. We leverage the advanced version of the recurrent neural network, that is, the long short-term memory network, to achieve the goal. We frame scoliosis as a classification problem and a regression problem. A network for classification is designed first. We perform the training and testing with real clinic records that are imputed by various tricks. Using this model, one can classify the current level of scoliosis into three predefined groups via a few publicly measurable indictors, such as body height or arm span. We also design a regression network that can predict the future progression of spine curvature. This model can infer the development in spine curvature at a certain time span according to the changes of other indictors. Both of these models are evaluated by various metrics. The experiment shows that the quantitative picture of the scoliosis can be captured by our models giving a significant performance boost. Hence, the resulting decision-support system can help to decide the necessity of a further intervene both for physicians and patients.

Xiaoyong Guo, Suxia Xu, Yizhong Wang, Jason Pui Yin Cheung, Yong Hu

A Hybrid Intelligent Framework for Thyroid Diagnosis

Thyroid disease exists across the whole world and many people are suffering from this disease. The diagnosis of thyroid disease is of great importance to human life. Although there are already some researches that introduces various methods for thyroid diagnosis and achieves good results, the performance of diagnosis still needs to be improved. Therefore, a hybrid intelligent framework, in which an optimal support vector machine (SVM) based on a hybrid optimization algorithm and a recursive feature elimination (RFE) method are incorporated, is proposed to predict thyroid disease in this paper. The hybrid optimization algorithm combines the teaching-learning based algorithm (TLBO) and differential evolution (DE), contributing to the parameter optimization of SVM. And the RFE method is introduced to obtain the optimal feature subsets for thyroid diagnosis. A thyroid dataset collected from UCI repository is utilized to evaluate the performance of the proposed framework. The experimental results demonstrate that the proposed framework achieves better and more stable performance than other compared methods.

Zhuang Li, Jingyan Qin, Xiaotong Zhang, Yadong Wan

Geriatric Disease Reasoning Based on Knowledge Graph

The lack of health care for ageing has become one of China’s most serious challgengs. The main work of this paper is building a database of a geriatric knowledge graph and proposing three inference rules based on Bayesian algorithm, which can effectively help the elderly to understand their health better and find out the abnormal condition as soon as possible. At the same time, it can assist doctors make auxiliary medical decisions and improve the cure rate. This article introduced a complete process of building a knowledge graph, from schema structure design to data acquisition, and processing the data until it fits the standard. Before applying to disease reasoning, we imported knowledge data into the Neo4j graph database to make full use of the inference flexibility and accuracy of the knowledge graph.

Shaobin Feng, Huansheng Ning, Shunkun Yang, Dongmei Zhao

Image Analysis Based System for Assessing Malaria

Malaria, a not only widespread but also a potentially fatal disease that can be found mainly in tropical regions of the world, with the World Health Organization reporting an estimated 219 million cases worldwide as of 2017 of which 435,000 were mortal. Diagnosis currently involves taking a blood sample from a patient who is presumed to be infected, which is examined under a microscope by trained experts, although reliable the process is tedious. This disease is an ever-increasing problem thereby creating a need for an automated solution to the diagnosis of malaria. The primary objective of this project is to design a tool that can diagnose malaria from an image of a blood smear that has been stained with the commonly used Giemsa stain (which highlights the parasites in a red blood cell by turning them dark purple). In this paper, we have developed a graphical user interface to assist with the separation of red blood cells and extraction of the cells infected with the malaria parasite as well as an ANN (Artificial Neural Network) for cell classification. The graphical user interface allows the user to analyse the blood sample by running a series of image processing techniques followed by the extraction of infected cells, the results have shown that these techniques could be potentially used to detect malaria. Currently, the achieved results shown that the proposed system has 92% accuracy of a database contains a large number of ground-truth images.

Kyle Manning, Xiaojun Zhai, Wangyang Yu

Research in Breast Cancer Imaging Diagnosis Based on Regularized LightGBM

Breast cancer is the main cause of cancer death in women, and it is increasingly threatening women’s health. The research of breast cancer imaging radiology aims to replace the traditional artificial diagnosis and promote the accurate diagnosis and treatment of breast cancer. Using deep learning technology to extract the features of breast images and to construct a breast cancer imaging diagnostic system is essential for promoting the development of diagnostic efficiency in the field of imaging medicine. Firstly, the existing mammography datasets are rotated 50 times at random angles (0°–360°) with the data enhancement technology. Then the images are encoded, and then the image features are extracted by ResNet50. In the process of classification, a regularized LightGBM model is added, the combination of the two models constitutes a model for breast cancer diagnosis. The addition of LightGBM improves the classification accuracy and performance of the model. The experimental results show that, the accuracy of the model on the two datasets of INbreast and DDSM is 91.7% and 93.6% respectively when doing the three-classification (normal/benign/malignant), when doing the binary classification (benign/malignant), the AUC on the two datasets are 0.942 and 0.962 respectively.

Chun Yang, Zhiguo Shi

Segmentation-Assisted Diagnosis of Pulmonary Nodule Recognition Based on Adaptive Particle Swarm Image Algorithm

The case characteristics of lung cancer are extremely complex, difficult to distinguish, and the rate of deterioration is rapid, and its early symptoms are not obvious. Early diagnosis and treatment of lung cancer is one of the main directions to reduce lung cancer mortality. The computer-aided detection and diagnosis system reduces the workload of the physician and improves the accuracy of image reading. In this paper, based on the analysis of the current lung nodule segmentation algorithm, in order to enhance the accuracy of lung nodule segmentation extraction, the adaptive particle swarm optimization algorithm is used to realize the simultaneous optimization of the number of mixed components and the model parameters, and finally realize the segmentation of lung nodules. The effectiveness and accuracy of segmentation of lung nodule recognition by adaptive particle swarm optimization algorithm is verified by adaptive particle swarm optimization and image model establishment. Provide new aids for the identification of pulmonary nodules.

Yixin Wang, Jinshun Ding, Weiqing Fang, Jian Cao

Auxiliary Recognition of Alzheimer’s Disease Based on Gaussian Probability Brain Image Segmentation Model

Alzheimer’s disease is an important disease that threatens the health of the elderly after cardiovascular disease, cerebrovascular disease and cancer. Early diagnosis and early intervention have an inestimable effect on disease control and treatment. Especially for China, which is facing the problem of population aging, early detection and early treatment are particularly important. According to the neuroimaging study of disease, by studying the degree of local brain loss in patients with Alzheimer’s disease, the disease information of the disease manifested in the brain structure is revealed, such as the decrease of the volume of the hippocampus and the thickness of the medial frontal temporal cortex. Thin and so on. In this paper, the local Gaussian probability image segmentation model is used to segment and extract the brain nuclear magnetic image, and the image segmentation of the hippocampus structure is extracted. The local Gaussian probability algorithm of image segmentation extraction algorithm is designed and optimized. The maximal posterior probability principle and Bayes’ rule are introduced to optimize the algorithm by grayscale processing of local image. Therefore, the Gaussian probability model is used to obtain the local mean and standard deviation as a function of spatial variation. Therefore, the probability model is more suitable for image segmentation with uneven gray scale than the probability model based on global hypothesis. Finally, experiments are carried out to verify the correctness of the theory and the robustness of Gaussian probability brain image segmentation.

Xinlei Chen, Dongming Zhao, Wei Zhong

Sleep Stage Classification Based on Heart Rate Variability and Cardiopulmonary Coupling

Sleep is a physiological process controlled by the autonomic nervous system. Autonomic activity differs in different stages of sleep. Heart Rate Variability (HRV) is a widely recognized indicator of autonomic activity and has been commonly used in sleep stage classification. However, HRV suffers from low repeatability and is very volatile. Cardiopulmonary Coupling (CPC) reflects autonomic activity from a different perspective and has been used to measure sleep quality. This paper explores the effect of using combination of HRV and CPC features in sleep stage classification. The experimental results using a decision-tree-based support vector machine (DTB-SVM) classifier on MIT-BIH polysomnographic database have shown that by adding three CPC features, the overall sleep stage classification accuracy has been raised from 95.74% (Kappa = 0.9257) to 96.89% (Kappa = 0.9449). The CPC features have shown to be superior in distinguishing deep sleep stage (with 3.69% increase). The classification accuracy of wake and light sleep has also improved with 1.67% and 0.89%, respectively.

Wangqilin Zhao, Xinghao Wu, Wendong Xiao

sEMG-Based Fatigue Detection for Mobile Phone Users

With the increasing widespread and popularity of internet connected smartphones, more and more people are becoming addicted to their mobile phones which has caused many health problems. Previous studies have proved that surface electromyographic (sEMG) signal can be used to monitor muscle fatigue in different situation such as driving environment or detect some cervical diseases such as muscle chronic pain. It inspired us an objective way to detect the fatigue status of phone users during a prolonged use of mobile phone. In this paper, an experiment was organized to collect phone users’ sEMG data and four classifiers were used with multiple sets of features for fatigue detection. Results show that the sEMG signal is an effective measure for detecting users’ neck fatigue, while the best classifier that achieved the highest accuracy compared to the other tested classifiers is the support vector machine (SVM).

Li Nie, Xiaozhen Ye, Shunkun Yang, Huansheng Ning

Research on the Effect of Video Games on College Students’ Concentration of Attention

Video games have become the main entertainment tool for people, occupying a lot of our time and attention, which is a limited resource and plays an important role in the organization and maintenance of intellectual activities. In the study, the influence of video games on college students’ attention was explored from two aspects (the game content and the time of game play) through the questionnaire survey and experimental measurement. A total of 310 persons participated in the survey, 27 of whom were involved in experiment. Results are given as follows. The more video games played, the harder attention is to focus. Also, non-violent video games have a negative short-term effect on players’ attention and reduced their attention level. But in the long term, non-violent video games can increase the females’ attention and reduce the males’ attention. Violent games affect players’ attention whether in the short term or in the long term. And playing games with high frequency will bring negative effect to players’ attention concentration, no matter non-violent or violent video games, no matter male or female players.

Zhixin Zhu, Jiansheng Li

Study on Cardiovascular Disease Screening Model Based-Ear Fold Crease Image Recognition

Cardiovascular disease screening is an effective means to effectively control the incidence of cardiovascular disease. The earlobe crease is an important marker for identifying cardiovascular diseases and can be used as an important sign for cardiovascular disease screening. Through the one-click uploading of human ear-based photos, the analysis of human ear crease based on image recognition is carried out, and the medical staff’s initial screening of cardiovascular disease assessment for some people is transformed into intelligent primary screening for cardiovascular disease in the city.

Xiaowei Zhong

Backmatter

Additional information

Premium Partner

    Image Credits