Sensor Data Analysis and Management
Description
Sensor Data Analysis and Management: The Role of Deep Learning delivers an insightful and practical overview of the applications of deep learning techniques to the analysis of sensor data. The book collects cutting-edge resources into a single collection designed to enlighten the reader on topics as varied as recent techniques for fault detection and classification in sensor data, the application of deep learning to Internet of Things sensors, and a case study on high-performance computer gathering and processing of sensor data.
The editors have curated a distinguished group of perceptive and concise papers that show the potential of deep learning as a powerful tool for solving complex modelling problems across a broad range of industries, including predictive maintenance, health monitoring, financial portfolio forecasting, and driver assistance.
The book contains real-time examples of analyzing sensor data using deep learning algorithms and a step-by-step approach for installing and training deep learning using the Python keras library. Readers will also benefit from the inclusion of:
* A thorough introduction to the Internet of Things for human activity recognition, based on wearable sensor data
* An exploration of the benefits of neural networks in real-time environmental sensor data analysis
* Practical discussions of supervised learning data representation, neural networks for predicting physical activity based on smartphone sensor data, and deep-learning analysis of location sensor data for human activity recognition
* An analysis of boosting with XGBoost for sensor data analysis
Perfect for industry practitioners and academics involved in deep learning and the analysis of sensor data, Sensor Data Analysis and Management: The Role of Deep Learning will also earn a place in the libraries of undergraduate and graduate students in data science and computer science programs.
More details
Other editions
Additional editions
Persons
A. Suresh, PhD is an Associate Professor in the Department of Computer Science and Engineering in SRM Institute of Science & Technology, Tamil Nadu, India. With nearly two decades of experience in teaching, his areas of specializations include Data Mining, Artificial Intelligence, Image Processing, Multimedia and System Software. He has two patents and has published approximately 90 papers in International journals. He is a Senior Member of IEEE, ISTE, MCSI, IACSIT, IAENG, MCSTA and a Global Member of Internet Society (ISOC). He has hosted two special sessions for IEEE sponsored conferences in Osaka, Japan and Thailand.
R. Udendhran is an Assistant Professor Grade III in the Department of Computer Science and Engineering, at the Sri SaiRam Institute of Technology, Chennai, India.
M. S. Irfan Ahmed is Associate Professor in the Department of Computer Science and Information, Faculty of Science and Literature at Taibah University, Saudi Arabia. He is a member of ISTE, MCSI, IACSIT, and IAENG.
Content
About the Editors vii
List of Contributors ix
Preface xiii
1 Efficient Resource Allocation Using Multilayer Neural Network in Cloud Environment 1
N. Vijayaraj, G. Uganya, M. Balasaraswathi, V. Sivasankaran, Radhika Baskar, and A.S. Syed Fiaz
2 Internet of Things for Human-Activity Recognition Based on Wearable Sensor Data 19
Dr. Vikram Rajpoot, Sudeep Ray Gaur, Aditya Patel, and Dr. Akash Saxena
3 Evaluation of Feature Selection Techniques in Intrusion Detection Systems Using Machine Learning Models in Wireless Ad Hoc Networks 33
T.J. Nagalakshmi, M. Balasaraswathi, V. Sivasankaran, D. Ravikumar, S. Joseph Gladwin, and S. Pravin Kumar
4 Neuro-Fuzzy-Based Bidirectional and Biobjective Reactive Routing Schema for Critical Wireless Sensor Networks 73
K.M. Karthick Raghunath and G.R. Anantha Raman
5 Feature Detection and Extraction Techniques for Real-Time Student Monitoring in Sensor Data Environments 97
Dr. V. Saravanan and Dr (Ms). N. Shanmuga Priya
6 Deep Learning Analysis of Location Sensor Data for Human-Activity Recognition 103
Hariprasath Manoharan, Ganesan Sivarajan, and Subramanian Srikrishna
7 A Quantum-Behaved Particle-Swarm-Optimization-Based KNN Classifier for Improving WSN Lifetime 117
Ajmi Nader, Helali Abdelhamid, and Mghaieth Ridha
8 Feature Detection and Extraction Techniques for Sensor Data 131
Dr. L. Priya, Ms. A. Sathya, and Dr. S. Thanga Revathi
9 Object Detection in Satellite Images Using Modified Pyramid Scene Parsing Networks 147
Akhilesh Vikas Kakade, S Rajkumar (Corresponding Author), K Suganthi, and L Ramanathan
10 Coronary Illness Prediction Using the AdaBoost Algorithm 161
G. Deivendran, S. Vishal Balaji, B. Paramasivan, S. Vimal (Corresponding Author)
11 Geographic Information Systems and Confidence Interval with Deep Learning Techniques for Traffic Management Systems in Smart Cities 173
Prisilla Jayanthi
Index 199
1
Efficient Resource Allocation Using Multilayer Neural Network in Cloud Environment
N. Vijayaraj1, G. Uganya 2, M. Balasaraswathi 3, V. Sivasankaran 4, Radhika Baskar 3, A.S. Syed Fiaz 1
1 Assistant Professor, CSE1 Assistant Professor, CSE, Vel Tech Rangarajan Dr. Sagunthala R & D Institute of Science & Technology, Chennai, Vel Tech Rangarajan Dr. Sagunthala R & D Institute of Science & Technology, Chennai2 Assistant Professor, ECE, Saveetha School of Engineering, SIMATS, Chennai
2 Assistant Professor, ECE3 Associate Professor, ECE, Saveetha School of Engineering, SIMATS, Chennai, Saveetha School of Engineering, SIMATS, Chennai4 Sreenivasa Institute of Technology and Management Studies, Chittoor, Andhra Pradesh
3 Associate Professor, ECE, Saveetha School of Engineering, SIMATS, Chennai
4 Sreenivasa Institute of Technology and Management Studies, Chittoor, Andhra Pradesh
1.1 Introduction
Nowadays, due to the development of the cloud computing environment, most information and communications technology (ICT) players have moved to new product management and application models-for example, Apple iCloud, Google App Engine, Amazon EC2, IBM Cloud, VMware Cloud, etc. Cloud computing is an important emerging field in ICT, making people's life easier by increasing productivity and processing speed, reducing cost and time consumption, facilitating backup and storing of multiple data, and enabling automation in the distribution of products and in development. It is quite challenging to offer trustworthy, powerful, qualitative, and cost-effective cloud services. In distributed cloud environments, a large number of dynamic resources are circulated around the world. Hence, the allocation of resources between the cloud user and cloud provider is complex, and cloud providers should be able to manage their resources with QoS and maximum customer satisfaction. Inadequate resource allocation leads to poor quality, bad performance, and substandard customer satisfaction, all falling below the criteria specified in service-level agreements (SLAs). Therefore, efficient and heterogeneous resource allocation is essential to avoid these problems.
In previous studies, the resource allocation problem was solved based on two methods: (i) reactive method, and (ii) proactive method. The reactive method is a common method in ICT, but it is not considered an effective method. The proactive method was developed to improve the performance of the system by allocating resources in a predefined manner. However, proactive-based methods, including time series (TS), queuing theory (QT), and reinforcement learning (RL), have some limitations. These include numerous data in TS, reconstruction of architecture when changing resources in QT, and large time requirements in RL. To resolve these proactive method constraints, feedback-based approaches have been introduced in difficult computing systems.
These feedback-based mechanisms are of two types, based on the allocation of resources to cloud services. These are (i) single input and single output (SISO), and (ii) multiple input and multiple output (MIMO). SISO is developed only for providing a single kind of resource allocation. However, in distributed cloud environments, users and providers need a mixture of resource allocation, and this leads to the violation of SLAs. To overcome this SISO limitation, the MIMO feedback control system was developed for multiple groupings of resource allocation by combining multiple numbers of separate SISO feedback control systems. But this type of MIMO feedback control system also leads to poor QoS and SLA violations. To overcome this issue, a coordinated multiple input multiple output feedback system was developed, which enhances the QoS by combining all the inputs in an integrated manner. Depending on the amount of work given to the services, the resources are allocated in a synchronized manner.
Other existing work focuses on adaptive multivariable resource mechanisms for multiple-type resource allocation with respect to cloud users and cloud providers. However, this has several limitations including scalability, reduced performance of allocated resources, and critical QoS.
To overcome these problems, we propose a better resource allocation system by using the RBNNOM multilayer neural network. Additionally, this novel method of RBNNOM considers priority-based resource allocation in terms of qualities and quantities and is responsible for QoS with good learning capabilities. Finally, this proposed work analyzes node prices and priority weights for cloud users and cloud providers in the cloud environment.
1.2 Related Work
Cloud computing is used in many applications including industrial, healthcare, environmental, and public science domains. However, nowadays, resource allocation is the key research area. Resource allocation is limited by factors such as bandwidth and bulk traffic. Yu et al. discussed a link-mapping algorithm for splitting the multiple paths and a migration algorithm [1] for relocation of paths in the virtual network (VN). This work is mainly proposed for reducing bandwidth requirements. Wei et al. proposed two methods for resource allocation based on game theory. These methods are binary integer (BI) programming method [2] and evolutionary mechanism (EM). In the first method, each and every node resolves the resource optimization problem separately, but this method is not suitable for multiple-resource tasks. Therefore, the second method has been designed additionally considering the multiplexing of resource assignments and is also used to reduce productivity loss. Yang et al. proposed a profile-based method for tackling the scalability problem in resource allocation [3]. The selection of profile is based on the application of the cloud environment. The resource is distributed in three ways. These are predistribution, distribution, and postdistribution.
Chen et al. designed a virtualization model and heuristic resource combination algorithm (HRCA) that is used to transfer physical type of resources to logical resources [4]. In this work, HRCA consists of two algorithms. These are matching and reconfiguration algorithms for allocating dynamic resources. Ramachandran et al. proposed two types of tenant models: the tenant requirement model (TRM) for measuring the functional requirements of tenants, and the tenant provider model (TPM) for allocating resources depending on dissimilar tenant information [5]. Wang et al. proposed a combined knowledge representation based on basic design, process, and product information [6]. In this work, the information structure is developed by four methods: filtering the information, summarizing the information of nodes, determining the group of nodes, and finding the solutions for difficult questions related to dynamic resource allocation.
Farahnakian et al. developed the ant colony system-virtual machine consolidation (ACS-VMS) for dynamic resource allocation [7]. This work is mainly used for reducing energy consumption when distributing resources to cloud users, as well as for reducing violations of SLA. Violations of SLA are mainly caused by migration and the overuse of resources. Sim proposed the focused selection contract net protocol (FSCNP) for the automatic collection of cloud services and service capability table (SCT) for recording the cloud services [8]. In this work, the author considers only the cost consumption, but it is not suitable for QoS and the cooperation of time slots.
Papagianni et al. discussed a unified resource allocation framework [9] that has two phases for the problem of mixed integer programming (MIP). These are the node mapping phase (NMP) and the link mapping phase (LMP), which provide a solution for the resource flow distribution problem. Linlin Wu et al. proposed a customer-driven SLA [10] that is mainly used for improving user satisfaction. This work is mainly used to reduce costs and SLA violations in software as a service (SaaS). However, this work needs further improvement in multitier applications. Tao et al. discussed the case library and Pareto-solution-based genetic algorithm (CLPS-GA) for heterogeneous resource allocation [11]. In this work, the authors mainly considered two types of programming information, including processor information and user requests. Peng et al. proposed a radial basis function neural network (RBFNN), based on multi-objective genetic algorithm [12], for optimum resource allocation. But, in this work, the authors did not consider the scalability, node price, and priority load.
In cloud environments, the allocation of resources is broadly classified into two methods: (i) reactive method, and (ii) proactive method. In reactive method, the resources are allocated after getting the threshold values. However, proactive method is preferred to reactive method because, in proactive method, resources are allocated in a predefined manner, which is used to mitigate traffic in complex systems.
1.2.1 Reactive Methods for Resource Allocation
Numerous methods are proposed to overcome the issues in resource allocation based on the reactive approach. Xiao et al. introduced virtual machine (VM) live migration technology for resource allocation depending on the applications. Also, the authors proposed the skewness method [13, 14], which is used to compute the equality of the resource utilization. But, in this...
