基於CSI 指紋庫之機器學習演算法做 帶缺陷訓練資料之室內定位;CSI-Based Fingerprints for Machine Learning Method with Indoor Localization of Incomplete Training Data

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Communication Engineering > Electronic Thesis & Dissertation > Item 987654321/86462

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/86462

Title:	基於CSI 指紋庫之機器學習演算法做帶缺陷訓練資料之室內定位;CSI-Based Fingerprints for Machine Learning Method with Indoor Localization of Incomplete Training Data
Authors:	許哲齊;Hsu, Che-Chi
Contributors:	通訊工程學系
Keywords:	通道狀態資訊;K-近鄰演算法;XGBoost;多層感知器;Channel State Information;K-Nearest Neighbors;Extreme Gradient Boosting;Multi-Layer Perception
Date:	2021-10-21
Issue Date:	2021-12-07 12:51:54 (UTC+8)
Publisher:	國立中央大學
Abstract:	室內定位(Indoor Localization) 是早於十餘年前就提出的技術，相較於全球定位系統(Global Positioning System, GPS) 的戶外定位系統(Location-Based Service, LBS)，由於衛星訊號無法穿透建材，故無法使用衛星進行定位。但室內定位的應用仍相當廣泛，且具有一定的商業價值，如醫療看護、倉儲、機場、百貨公司等，加上近年來科技技術的提升，使得室內定位搖身一變成為一個熱門的話題。本篇論文採用Wi-Fi 做為室內定位技術，以通道狀態資訊(Channel State Information, CSI) 作為定位方針，先敘述環境參數與設備，並說明各式演算法，從而討論利用各種演算法來進行基本室內定位的準確度，再延伸討論至若有些位置無法採集或不方便採集到通道資訊時，能以什麼樣的方式來進行克服，進而在資訊存有缺陷的情況下實現室內定位。;Indoor Localization has been being an attracting technology studied over ten years along with the Location-Based Service of Global Positioning System (GPS). Because satellite signals cannot enter building, it is impossible to use satellite signals for indoor localization. However, the application of indoor localization is still quite extensive, and it has some commercial values certainly, such as medical care, warehousing, airports, departments, etc. In recent years, due to the advancement of the computing technology, the issue of indoor localization become an attractive topic. This thesis adopts Wi-Fi as the technique of Indoor Localization and channel state information (CSI) as the localization fingerprint. First, we describe the problem formulation and the involved equipment. Second, we introduce some machine learning algorithms that are compared in this paper. Then we discuss the accuracy of indoor localization with different machine learning algorithms. In this paper, considering there is some missed information which cannot be collected or is inconvenient to be collected for machine training, we propose a method that can be used to deal with this kind of situation to achieve effective indoor localization.
Appears in Collections:	[Graduate Institute of Communication Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	146	View/Open

社群 sharing

Loading...