CHOSUN

Practical Indoor Localization System Using Bluetooth Low Energy Beacons

Metadata Downloads
Author(s)
수베디 산토쉬
Issued Date
2019
Abstract
In recent time, social and commercial interest on location-based service (LBS) is significantly increased owing to the rise in the number of smart devices and technologies. The global navigation satellite systems (GNSS) has long been employed for LBS to navigate and determine accurate and reliable location information at outdoor environments. However, the GNSS signals are too weak to penetrate buildings and unable to provide reliable indoor LBS. Hence, the incompetence of GNSS at indoor environment invites extensive research and development of an indoor positioning system (IPS).
Various technologies and techniques have been studied for IPS development. The radio frequency based wireless technologies, particularly Wi-Fi and Bluetooth low energy (BLE) are widely used for indoor LBS. The localization techniques on IPS are proximity, trilateration, triangulation, and fingerprinting localization. The fingerprinting localization method is largely accepted for IPS development owing to its good localization accuracy. The signal measurement principles like the angle of arrival (AOA), time of arrival (TOA), time difference of arrival (TDOA), and received signal strength (RSS) are adopted for realizing the localization techniques. The wireless technology, the positioning technique, and the signal measurement principle can vary according to the required performance metrics of the IPS.
In this dissertation, a practical IPS is presented to solve the real issues of conventional fingerprinting localization. The first half of the dissertation is dedicated to IPS problems, indoor localization fundamentals, and the AP signal characteristics and filtration. First, the hurdles behind the realization of practical IPS is discussed with different localization techniques and signal measurement principles. Second, the basic radio frequency (RF) based IPS technologies and their signal measurement principles are elucidated. In addition, a brief discussion of different positioning algorithms and the performance metrics of the IPS are presented. Third, the dissertation presents the RSS characteristics of Wi-Fi and BLE access points (APs) and estimates the signal propagation parameters of both the wireless technologies in the testbed environment. Moreover, some signal filtering methods are elaborated to smooth the time-varying and fluctuating RSS.
Although fingerprinting is widely adopted for designing indoor positioning system (IPS), it holds significant problems. The labor-intensive and time-consuming offline phase is the major issue in the realization of fingerprinting localization. In addition, computational complexity can possess an adverse effect on localization efficiency. At first, the conventional weighted k-nearest neighbor (Wk-NN) fingerprinting localization is illustrated. Two different improvisation techniques over the typical Wk-NN fingerprinting localization are proposed in the dissertation. The first suggested technique combines the Wk-NN fingerprinting with the weighted centroid (WC) localization to reduce the number of reference points (RPs) for acceptable localization accuracy. Since the number of RPs are reduced, it reduces the time required for acquiring fingerprinting data. Similarly, the second approach employs multiple features to represent a fingerprinting of an RP and utilizes clustering to reduce the computational complexity. This approach uses the WC, a rank and signal strength of APs as the fingerprint information. It uses affinity propagation clustering (APC) as a clustering component that reduces the computational time and the localization estimation error of the IPS.
While the probabilistic approach of fingerprinting yields more accurate localization result than the Wk-NN fingerprinting, it is computationally expensive. Moreover, the time-consuming and labor-intensive offline workload intensify the practical limits to realize a practical IPS. Hence, keeping this fact on the mind, a machine learning approach of fingerprinting localization is suggested along with the APC, which helps to reduce both the offline workload and online computational cost. The proposed approach predicts the fingerprint information at locations with no prior measurements and clusters the predicted RSS information using Gaussian process regression (GPR) and APC, respectively.
The IPS approaches on this dissertation rely on the received signal fingerprint of BLE beacons that is easy to acquire with the modern smartphones. It is expected that the proposed method will be beneficial in various LBSs.|최근 들어 스마트 기기·기술의 향상으로 위치기반 서비스(LBS)에 대한 사회적·상업적 관심이 크게 높아지고 있다. 이를 위해 GNSS(Global Navigation Satelite System)이 실외 환경에서 정확하고 신뢰할 수 있는 위치 정보를 탐색하고 결정하기 위해 오랫동안 사용되어 왔다. 그러나 GNSS 신호는 너무 약해서 건물을 통과할 수 없고 신뢰할 수 있는 실내 LBS를 제공할 수 없으므로 실내 측위 시스템(IPS)에 대해서 광범위한 연구 개발을 시작하게 되었다.
PS 개발을 위해 다양한 기술과 기법이 연구되어 왔다. 무선 주파수 기반 무선기술, 특히 Wi-Fi와 BLE(Bluetooth Low Energy)는 실내 LBS에 널리 사용된다. IPS를 구현하기 위한 기술로는 근접성, 삼각측량, 삼변측량, 핑거프린팅 등이 있다. 핑거프린팅 기법은 정확도가 뛰어나서 IPS 개발에 널리 적용되어지고 있다. 위치 정보를 얻어내기 위해 AOA(Angle of arrival), TOA(Time of arrival), TDOA(Time difference of arrival), RSS(Received signal strength)와 같은 신호 측정 방법이 활용되고 있으며, 무선기술, 위치 결정 및 신호 측정 방식에 따라 측위 정확도는 달라질 수 있다.
본 논문에서는, 기존의 핑거프린팅 기반 측위 기술의 실질적인 문제를 해결하기 위한 측위 기술들을 제시한다. 첫째, IPS 문제, 실내 위치 측위 기초, AP 신호 특성 및 여과 등을 설명한다. 또한, 위치 측위의 다양한 기법과 신호 측정 원리로 실용 IPS의 실현 이면의 장애물에 대해 논의한다. 둘째, 기본적인 무선 주파수(RF) 기반 IPS 기술과 그 신호 측정 원리를 설명한다. 또한, IPS를 위한 다양한 위치 결정 알고리즘과 성능 지표를 간략히 논한다. 셋째, Wi-Fi와 BLE 접속점(AP)의 RSS 특성을 제시하고, 테스트베드 환경에서 두 무선기술의 신호 전파 매개변수를 추정한다. 그리고, 시간 변동과 변동하는 RSS를 원활하게 하도록 몇 가지 신호 필터링 방법을 제시한다.
실내 위치 확인 시스템(IPS) 설계에 핑거프린팅 기법이 널리 채택되고 있지만 상당한 문제를 가지고 있다. 계산 복잡성은 측위 효율에 부정적인 영향을 미칠 수 있다. 본 논문에서 일반적인 Wk-NN 핑거프린팅과 WC(Weighted Centroid)를 결합하여 허용 가능한 측위 정확도를 위해 RP(Reference Point)의 수를 줄인다. RP 수가 줄어들기 때문에 핑거프린팅 데이터에 필요한 시간을 단축한다. 마찬가지로, 두 번째 접근방식은 RP의 핑거프린팅을 나타내기 위해 여러 가지 특징을 채택하고, 계산 복잡성을 줄이기 위해 클러스터링을 이용한다. 이 접근방식은 AP의 등급 및 신호 강도인 WC를 핑거프린팅으로 사용한다. IPS의 계산 시간과 측위 추정 오류를 줄이는 클러스터링 구성요소로 APC(Affinity Propagation Clustering)를 사용한다.
비록 핑거프린팅 확률론적 접근방식이 Wk-NN 핑거프린팅 보다 더 정확한 결과를 산출하지만, 측위 효율성은 떨어진다. 시간이 오래 걸리고 노동 집약적인 오프라인 작업량은 실용적 IPS를 실현하기 위해 핑거프린팅 기반 기계 학습 접근법을 제안한다. 사전 측정 없이 핑거프린팅 정보를 예측하고 GPR(Gaussian Process Regression) 및 AP본 논문에서는 IPS를 위해 스마트폰과 BLE 비콘의 수신 신호 핑거프린팅을 이용한다. 제안된 방법이 다양한 LBS에서 유익할 것으로 기대한다.
Alternative Title
저전력 블루투스 비콘을 활용한 실용성 있는 실내 측위 시스템
Alternative Author(s)
Santosh Subedi
Department
일반대학원 정보통신공학과
Advisor
변재영
Awarded Date
2019-08
Table Of Contents
Abstract Korean i
Abstract iii
Acronyms vi
Contents vii
List of Figures x
List of Tables xiii
1 INTRODUCTION 1
1.1 Location-based Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The IPS Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Research Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Contributions of Dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.5 Thesis Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 INDOOR LOCALIZATION FUNDAMENTALS 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 The Technologies for Indoor Localization . . . . . . . . . . . . . . . . . . . . . 5
2.2.1 RFID/NFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.2 UWB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.3 WSN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.4 Wi-Fi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.5 BLE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Signal Measurement Principles . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3.1 RSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3.2 TOA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.3 TDOA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.4 AOA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Algorithms for IPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4.1 Proximity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4.2 Lateral/Angular . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.3 Fingerprinting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.4 Dead-reckoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 The Performance Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5.1 Accuracy and Precision . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5.2 Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5.3 Scalability and Robustness . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5.4 Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3 AP SIGNAL CHARACTERISTICS AND FILTRATION 24
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 RSS characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 RSS Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3.1 Averaging/Moving Average Filter . . . . . . . . . . . . . . . . . . . . . 31
3.3.2 Kalman Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3.3 Gaussian Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3.4 Exponential averaging (EA) filter . . . . . . . . . . . . . . . . . . . . . 36
3.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4 IMPROVISATION OVER CONVENTIONAL FINGERPRINTING LOCALIZATION
39
4.1 Process Flow of Wk-NN Fingerprinting Localization . . . . . . . . . . . . . . . 39
4.1.1 Offline or Training Phase . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.1.2 Online or Execution Phase . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.1 Wk-NN Fingerprinting Approaches . . . . . . . . . . . . . . . . . . . . 42
4.2.2 Hybrid Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.3 Clustering Based Approaches . . . . . . . . . . . . . . . . . . . . . . . 43
4.3 Proposed Method-I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.3.1 Process Flow of the Proposed Positioning Approach . . . . . . . . . . . 44
4.3.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.3 Experimental Results and Discussion . . . . . . . . . . . . . . . . . . . 51
4.4 Proposed Method-II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.4.1 Affinity Propagation Clustering . . . . . . . . . . . . . . . . . . . . . . 62
4.4.2 Process Flow of the Proposed Method . . . . . . . . . . . . . . . . . . . 64
4.4.3 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.4.4 Experimental Result and Discussion . . . . . . . . . . . . . . . . . . . . 69
4.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5 THE PRACTICAL FINGERPRINTING LOCALIZATION 77
5.1 Probabilistic Approach of Fingerprinting Localization . . . . . . . . . . . . . . . 77
5.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.3.1 Offline Workload Reduction without Regression . . . . . . . . . . . . . 79
5.3.2 Offline Workload Reduction with Regression . . . . . . . . . . . . . . . 79
5.3.3 Computational complexity reduction with APC . . . . . . . . . . . . . . 80
5.4 Gaussian Process Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.5 Proposed Practical Fingerprinting Method . . . . . . . . . . . . . . . . . . . . . 82
5.5.1 Process Flow of the Proposed Practical Positioning Approach . . . . . . 82
5.5.2 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.5.3 Experimental Results and Discussion . . . . . . . . . . . . . . . . . . . 86
5.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6 CONCLUSIONS 93
Degree
Doctor
Publisher
조선대학교 일반대학원
Citation
수베디 산토쉬. (2019). Practical Indoor Localization System Using Bluetooth Low Energy Beacons.
Type
Dissertation
URI
https://oak.chosun.ac.kr/handle/2020.oak/13922
http://chosun.dcollection.net/common/orgView/200000267387
Appears in Collections:
General Graduate School > 4. Theses(Ph.D)
Authorize & License
  • AuthorizeOpen
  • Embargo2019-08-23
Files in This Item:

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.