Ultrawideband based Localization for Harsh Environments Jiwoong Park
- 주제(키워드) UWB Localization , Autonomous Vehicle , Online Learning , NLOS identification
- 주제(DDC) 004.6
- 발행기관 아주대학교 일반대학원
- 지도교수 Young-Bae Ko
- 발행년도 2024
- 학위수여년월 2024. 2
- 학위명 박사
- 학과 및 전공 일반대학원 AI융합네트워크학과
- 실제URI http://www.dcollection.net/handler/ajou/000000033614
- 본문언어 영어
- 저작권 아주대학교 논문은 저작권에 의해 보호받습니다.
초록/요약
With the recent deployment of Ultrawideband (UWB) in smartphones, the demand for UWB localization has increased significantly. Since localization has been a well-studied technology for a long time, many studies proposed to adapt existing localization techniques to UWB localization. However, these solutions do not account for the environment-sensitive nature of UWB. They required a controlled environment to provide high-accuracy localization or could provide low-accuracy localization with meter-level errors in a harsh environment. This dissertation focuses on UWB localization techniques that can provide accurate UWB localization in a harsh environment. To achieve these goals, we first determined the two most important scenarios for UWB localization: 1) vehicular scenario and 2) indoor scenario. We propose UWB localization techniques to address the challenges of harsh environments in our target scenarios. At first, we propose a UWB localization scheme for vehicular scenarios, including a novel ranging protocol and an innovative peak detection-based receiving method. After that, we present a deep learning-based NLOS identification algorithm and an online model update algorithm to solve the problem of unpredictable NLOS in indoor scenarios. This dissertation provides an intensive performance evaluation of commercial UWB devices in practical environments.
more목차
I. Introduction 1
1.1 Motivation 1
1.2 Contributions of the Dissertation 4
1.3 Overview of the Dissertation 5
II. Backgrounds and Related Works 7
2.1 UWB Wireless Communication 8
2.1.1 An Overview of the IEEE 802.15.4a Standard 9
2.1.2 An Overview of the IEEE 802.15.4z Standard 12
2.1.3 Other Standards 13
2.2 UWB Localization 15
2.2.1 Why is UWB Ranging accurate 15
2.2.2 UWB Ranging Protocols 17
2.2.3 UWB Localization Algorithms 21
2.2.4 Major Sources of UWB Localization Error 23
2.2.5 UWB Channel Impulse Response 25
2.3 Related Works 27
III. UWB based Pedestrian Localization for Autonomous Vehicles 31
3.1 Introduction 32
3.2 Related Works 37
3.3 PedLoc System Design 41
3.3.1 An Overview of PedLoc 41
3.3.2 Step 1: Root Anchor Selection 42
3.3.3 UWB Channel Quality 44
3.3.4 Step 2: Root Anchor Ranging 46
3.3.5 Step 3: Overhearing 47
3.3.6 Peak Detection Algorithm 49
3.3.7 PedLoc Localization Algorithm 52
3.4 Implementation and Evaluation 55
3.4.1 Experimental Methodology 55
3.4.2 Performance Evaluation 59
3.4.3 Discussing Core Components 65
3.4.4 Summary of the Chapter 70
IV. Online Deep Learning based NLOS Identification in Unfamiliar Indoor Environments 71
4.1 Introduction 72
4.2 Related Works 75
4.2.1 Algorithm-based approaches 75
4.2.2 Machine Learning-based approaches 76
4.3 Problem Statement 80
4.4 Proposed Scheme 83
4.4.1 Deep Learning Model 84
4.4.2 Transfer Learning 87
4.4.3 Online Update Algorithm 89
4.5 Evaluation 92
4.5.1 Experimental Methodology 92
4.5.2 Performance Evaluation 93
4.6 Summary of the Chapter 102
V. Conclusion and Future Works 103
References 105
