검색 상세

IoT환경에서의 부하 균형과 신뢰성을 위한 클러스터링 기법 연구

Clustering algorithms for Load Balancing and Reliability in IoT

초록/요약

Recently, many researches focus on Wireless Sensor Networks (WSN) which consist of low-cost, low power, and short RF range sensor devices. Energy efficiency and traffic load balancing is a key challenge in design of WSN. Clustering techniques can be an efficient solution to solve energy and load balancing issues in WSN. Furthermore, several applications requires specific delay parameters to satisfy the requirements. In order to satisfy these requirements, the network designer have to consider the guaranteed delay for these applications. In this thesis, we introduce entropy based network clustering algorithm to solve energy and delay issues in WSN. Our algorithm uses an entropy based Cluster Cost function which considers traffic load and traffic variation. Traditionally, industrial networks are constructed through wired communications. Wired networks require expensive electronic cable and maintenance costs. With recent advances in wireless sensor networks (WSNs), the realization of low-cost Industrial Wireless Sensor Network (IWSN) has become feasible. Compare to wired networks, IWSN has several constraints like resource, reliability, redundancy and so on. These constraints impose many challenges for designing routing protocols of ISWN. Energy efficiency and reliability are important issues, in most case there is a trade-off between energy efficiency and reliability. Existing IWSN standards like ISA100.11a and WirelessHART propose simple and reliable graph routing, while the details of graph generation scheme are not specified in the standard. We hope that routing protocols decrease energy consumption while the link reliability can be guaranteed. In this paper, we propose a novel graph generation scheme which can enhance communication reliability with low energy consumption. The proposed scheme provides entropy-based clustering and multi-path graph generation algorithms to satisfy the industrial application’s reliability requirement.

more

목차

1. Introduction
1.1 Motivation
1.2 Thesis Objective
1.3 Contribution
1.4 Structure of Thesis
2. Background
2.1 Clustering Overview
2.2 Information Entropy
2.3 Industrial WSN : ISA 100.11a
2.3.1 Overview of ISA 100.11a
3. An Entropy Based Clustering Algorithm
3.1 Overview
3.2 Entropy Based Clustering Algorithm
3.2.1 Energy Model
3.2.2 Information Entropy Modeling
3.2.3 CH Election Phase in Clustering Algorithm
3.2.4 Cluster Formation Phase in Clustering Algorithm
4. Cluster based Graph Generation Scheme
4.1 Overview
4.2 Cluster based Graph Generation Scheme
4.2.1 Entropy Based Clustering
4.2.2 Reliable Graph Generation
5. Performance Evaluation
5.1 Entropy based Clustering Algorithm Performance Analysis
5.2 Cluster based Graph Generation Algorithm Performance Analysis
6. Conclusion
6.1 Conclusion
6.2 Open problem & Further work
REFERENCE

more