Work

An Indoor-Oriented Localization And Navigation System

Year: 2019
Type: article
Abstract: QIWEI BAODue to the advancing demand for user real-time tracking and routing inside buildings, several technologies are proposed for indoor localization and navigation services.Wi-Fi fingerprint positioning usually owns a poor stability because of the impact of complex surroundings inside buildings, and PDR (Pedestrian Dead Reckoning) has accumulating error.For architecture with complicated indoor environment, an algorithm for shortest path finding is required.An indoor-oriented localization and navigation system is presented in this thesis with a hybrid user's localization approach and an optimized shortest path finding algorithm for navigation.The system first estimates user's position by calculating the direction and stride of each step.A Wi-Fi RSSI (Received Signal Strength Indicator) fingerprint method is implemented in the meanwhile collecting user's location information.After that, K-Nearest Neighbor (KNN) algorithm is employed on the RSSI data as a feedback correction process.The system is designed for Android mobile phones so the users can take it for navigation in a large building.Experimental results show that the proposed approach can achieve an overall location error of about 1.67 meters of 20 meters user's walking distance (Accuracy: 91.65%).x 3).To reduce the accumulated errors from the detection of each step, a KNN-based Wi-Fi RSSI positioning method is presented. Thesis OrganizationIn Chapter 2, several existing technologies suitable for indoor positioning are reviewed.A literature survey is presented comparing several systems using these technologies.Chapter 3 illustrates the proposed approach within a flow diagram description of the progress from system initialization to user arrival.Chapter 4 presents two main factors in the location database: 1).how to define base stations and sampling points on the floor plans and 2).what to be included in the location database.In Chapter 5, a PDR (Pedestrian Dead Reckoning) method is proposed to estimate user's position.And a series of filtering algorithms is used to improve the accuracy of stride estimation.
Author: Bao, Qiwei
Language: English
Open Access status: green
OpenAlex ID: W7019774073
FWCI: 0.09
Cites: 22
Cited by: 1
Related to: 10