Collection Title:	SIU Thesis
Title :	Optimization Approaches to Cost Effective Resource Management for Cloud-based IoT Applications
Material Type:	printed text
Authors:	Nay Myo Sandar, Author ; Myint, Lin Min Min, Associated Name ; Chaisiri Sivadon, Associated Name
Publisher:	กรุงเทพฯ: มหาวิทยาลัยชินวัตร
Publication Date:	2017
Pagination:	x, 124 p.
Layout:	ill, tables
Size:	30 cm.
Price:	500.00
General note:	SIU THE: SOST-PhD-IT-2017-02 Thesis. [PhD [Information Technology]]. -- Shinawatra University, 2017
Languages :	English (eng)
Descriptors:	[LCSH]Internet of Things [LCSH]Wireless Sensor Networks
Keywords:	Internet of Things (IoT), Wireless Sensor Network (WSN), Cloud Computing, Aggregator Approach, Software Defined Network Approach, Optimization Approaches
Abstract:	Over the last decade, the growing popularity of various devices are increasingly connected to the Internet and led to the vision of Internet of Things (IoT). In IoT, wireless sensor networks (WSNs) deploy sensors to gather data around their surroundings. Nowadays, Cloud computing can support scalable storage and processing task for large sensor data. However, sensors can encounter with bandwidth constraint and memory constraint to directly transfer data to the cloud. This thesis proposes a framework using aggregator approach and software defined network (SDN) approach for cloud-based IoT applications. The aggregator approach provides aggregators to buffer data from sensors and forward to the cloud. The SDN approach provides scalable bandwidth and programmable network paths to deliver data between aggregators and cloud. Applying aggregator and SDN approaches in the framework can achieve the expected seamless network connectivity between numerous sensors and cloud. Since the proposed framework is designed by aggregator approach, SDN approach, and cloud computing, we mainly focus on achieving cost effectiveness for capacity planning of aggregators, provisioning of SDN bandwidth and cloud resources. The capacity planning of aggregators is a critical issue to purchase the optimal number of aggregators from third party providers for handling data from multiple sensors while the investment is minimized. Then, Internet Service Provider (ISP) provides SDN bandwidth and cloud providers provide cloud resources with reservation and on-demand options. The on-demand option allows consumers to dynamically provision resources anytime but it leads to higher cost. The reservation option is cheaper but consumers must prepay specific resources for long term plan before the demand is known. Without knowing demand, provisioning resources with reservation option could experience underprovisioning or overprovisioning. To tackle resource management problems, this thesis proposes several algorithms in three use cases of cloud-based IoT applications, i.e., video monitoring or surveillance system, meteorological monitoring system, and healthcare monitoring system. First, an optimal on-demand cloud resource provisioning algorithm is proposed for video monitoring or surveillance system. This algorithm applies binary integer programming to allocate video streams from cameras among cloud providers with on-demand option. The numerical results can choose optimal cloud providers where video streams are allocated with the minimum cost. Second, a joint resource management algorithm is proposed for meteorological monitoring system. This algorithm applies deterministic multi-server queuing theory for capacity planning of aggregators to provide sufficient service to data from meteorological sensors while reducing the investment and deterministic integer programming to provision the precise amount of cloud resources with reservation option based on data demand certainty for stationary sensors. The numerical results can determine optimal number of aggregators and provision optimal amount of cloud resources with the minimum total cost. Finally, a unified resource management algorithm is proposed for patients’ healthcare monitoring in hospital. This algorithm applies Markov multi-server queuing theory for capacity planning of aggregators to efficiently operate data from body sensors attached to patients within hospital while investment is reduced and two stage stochastic programming to achieve the best advance reservation of SDN bandwidth and cloud resources under data demand uncertainty for mobile sensors. The numerical results can provide the minimum total cost by allocating optimal number of aggregators, amount of SDN bandwidth and cloud resources. As aforementioned, several algorithms are proposed to minimize the total cost for resource management problems in three usage scenarios of cloud-based IoT applications. In practice, the proposed algorithms can be applicable with the minimum total cost for any cloud-based IoT applications.
Curricular :	BSCS/MSIT/PhDT
Record link:	http://libsearch.siu.ac.th/siu/opac_css/index.php?lvl=notice_display&id=27301

SIU Thesis. Optimization Approaches to Cost Effective Resource Management for Cloud-based IoT Applications [printed text] / Nay Myo Sandar, Author ; Myint, Lin Min Min, Associated Name ; Chaisiri Sivadon, Associated Name . - [S.l.] : กรุงเทพฯ: มหาวิทยาลัยชินวัตร, 2017 . - x, 124 p. : ill, tables ; 30 cm.
500.00
SIU THE: SOST-PhD-IT-2017-02
Thesis. [PhD [Information Technology]]. -- Shinawatra University, 2017
Languages : English (eng)

Descriptors:	[LCSH]Internet of Things [LCSH]Wireless Sensor Networks
Keywords:	Internet of Things (IoT), Wireless Sensor Network (WSN), Cloud Computing, Aggregator Approach, Software Defined Network Approach, Optimization Approaches
Abstract:	Over the last decade, the growing popularity of various devices are increasingly connected to the Internet and led to the vision of Internet of Things (IoT). In IoT, wireless sensor networks (WSNs) deploy sensors to gather data around their surroundings. Nowadays, Cloud computing can support scalable storage and processing task for large sensor data. However, sensors can encounter with bandwidth constraint and memory constraint to directly transfer data to the cloud. This thesis proposes a framework using aggregator approach and software defined network (SDN) approach for cloud-based IoT applications. The aggregator approach provides aggregators to buffer data from sensors and forward to the cloud. The SDN approach provides scalable bandwidth and programmable network paths to deliver data between aggregators and cloud. Applying aggregator and SDN approaches in the framework can achieve the expected seamless network connectivity between numerous sensors and cloud. Since the proposed framework is designed by aggregator approach, SDN approach, and cloud computing, we mainly focus on achieving cost effectiveness for capacity planning of aggregators, provisioning of SDN bandwidth and cloud resources. The capacity planning of aggregators is a critical issue to purchase the optimal number of aggregators from third party providers for handling data from multiple sensors while the investment is minimized. Then, Internet Service Provider (ISP) provides SDN bandwidth and cloud providers provide cloud resources with reservation and on-demand options. The on-demand option allows consumers to dynamically provision resources anytime but it leads to higher cost. The reservation option is cheaper but consumers must prepay specific resources for long term plan before the demand is known. Without knowing demand, provisioning resources with reservation option could experience underprovisioning or overprovisioning. To tackle resource management problems, this thesis proposes several algorithms in three use cases of cloud-based IoT applications, i.e., video monitoring or surveillance system, meteorological monitoring system, and healthcare monitoring system. First, an optimal on-demand cloud resource provisioning algorithm is proposed for video monitoring or surveillance system. This algorithm applies binary integer programming to allocate video streams from cameras among cloud providers with on-demand option. The numerical results can choose optimal cloud providers where video streams are allocated with the minimum cost. Second, a joint resource management algorithm is proposed for meteorological monitoring system. This algorithm applies deterministic multi-server queuing theory for capacity planning of aggregators to provide sufficient service to data from meteorological sensors while reducing the investment and deterministic integer programming to provision the precise amount of cloud resources with reservation option based on data demand certainty for stationary sensors. The numerical results can determine optimal number of aggregators and provision optimal amount of cloud resources with the minimum total cost. Finally, a unified resource management algorithm is proposed for patients’ healthcare monitoring in hospital. This algorithm applies Markov multi-server queuing theory for capacity planning of aggregators to efficiently operate data from body sensors attached to patients within hospital while investment is reduced and two stage stochastic programming to achieve the best advance reservation of SDN bandwidth and cloud resources under data demand uncertainty for mobile sensors. The numerical results can provide the minimum total cost by allocating optimal number of aggregators, amount of SDN bandwidth and cloud resources. As aforementioned, several algorithms are proposed to minimize the total cost for resource management problems in three usage scenarios of cloud-based IoT applications. In practice, the proposed algorithms can be applicable with the minimum total cost for any cloud-based IoT applications.
Curricular :	BSCS/MSIT/PhDT
Record link:	http://libsearch.siu.ac.th/siu/opac_css/index.php?lvl=notice_display&id=27301