| Author: | Mostafa, Ahmed Mansour Abdalla |
| Title: | Indoor localization based on multi-sensor fusion, crowdsourcing, and multi-user collaboration |
| Advisors: | Chen, Wu (LSGI) |
| Degree: | Ph.D. |
| Year: | 2023 |
| Subject: | Indoor positioning systems (Wireless localization) Location-based services Mobile geographic information systems Mobile computing Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Land Surveying and Geo-Informatics |
| Pages: | xviii, 171 pages : color illustrations |
| Language: | English |
| Abstract: | In enclosed spaces such as indoor and deep urban environments, the omnipresent signals of Global Navigation Satellite Systems (GNSS) cannot provide accurate localization services because of signal attenuation and blocking. Unfortunately, these areas are where people spend most of their time. Moreover, accurate indoor positioning is tremendously required for easy navigation in modern buildings since urbanization grows the buildings' complexity and size. Furthermore, indoor localization is required to provide lifesaving services such as surveillance programs and firefighters. In addition to helping save lives, indoor localization is needed to improve business operations. Generally, in the current era, Location-based services (LBS) are in continuous demand at any time and in all environments. Indeed, the lack of accurate localization in indoor spaces and when shuttling from outdoors to indoors hinders ubiquitous and seamless LBS. Fortunately, sensors embedded in current smartphones eliminate the cost barrier of end-user devices to develop different localization techniques and adopt multi-sensory integration, crowdsourcing, and multi-user collaboration approaches for localization purposes. On this ground, this research aims to develop a low-cost, self-deployable, and ubiquitous Indoor Positioning System (IPS) using off-the-shelf smartphone sensors and pervasive signals. To achieve that, an enhanced indoor-outdoor (I/O) detection service is first proposed to autonomously distinguish the type of ambient environment. The contribution of the proposed service is as follows: it utilizes indicators of Low Power Consumption Sensors (LPCS) to execute continuous detection tasks, invoke the GNSS in the confusion scenarios and transition intervals to give most firm decision on the credibility of the transition triggered by LPCS, and compensate thresholds of light indicators. In this manner, GNSS is used for short intervals that help reduce detection latency and power consumption, besides ensuring accurate and reliable detection. The proposed I/O detection uses to realize seamless and user-friendly navigation by automatically invoking the localization sensors for the distinguished environment, thereby reducing superfluous Wi-Fi scanning outdoors and deactivating GNSS indoors. In the indoor environments, which is the focus of this study, smartphone richness with multiple sensors can be exploited to observe pervasive indoor signals, such as Wi-Fi-based Received Signal Strength (RSS) and Magnetic Field (MF). Wireless localization based on fingerprinting techniques is preferred to leverage the pervasive wireless signals because fingerprinting mitigates the multipath effect. Offline fingerprinting databases, however, require manual training and updating, which is time-consuming, labor-intensive, and limits their scalability. By autonomously developing these offline databases from pervasive resources, self-deployable systems can be achieved. On this basis, this study proposes a framework “Everywhere” to leverage pervasive crowdsourced data to develop a ubiquitous IPS. Compared to counterparts' studies, the proposed framework makes the following contributions: 1) to reduce the cost borne by the user device and ensure widespread adoption of crowdsourcing, a strategy is introduced to manage the collection process of crowdsourced data. This strategy takes advantage of a) the proposed I/O detection service to confine the data collection to GNSS-denied areas only; and b) the database availability and motion mode to manage the collection rate; 2) selection criteria are introduced to qualify inertial data. These criteria depend mainly on the characteristics of the collected data and do not rely on factors that require the deployment of internal auxiliary Anchor Nodes (ANs) or floor plans (i.e., as suggested in previous studies), thereby maintaining the system's ubiquity; 3) the large errors expected while depending on GNSS ANs to localize crowdsourced signatures in buildings surrounded by GNSS-denied areas or multistory buildings are mitigated by proposing leveraging the elevators as a source of detecting or deploying AN. The proposed location provides a trade-off solution that maintains a high localization accuracy at all floors with low deployment cost and effort. Precisely, AN can be detected at all floors with cost minimized by a factor of N:1, where N is the number of floors. ANs such as BLE beacons can be effortlessly deployed in existing building elevators or incorporated into elevators as a pre-installed component, similar to security cameras; 4) the data accumulation over time is proposed to autonomously derive the identifiers, locations, and propagation information of fixed Wi-Fi APs to act as internal ANs and eliminate the need to deploy auxiliary ANs at each floor. Inferences from these pervasive ANs help align the traces collected entirely within the floor area, thereby improving trace localization and extending the spatial coverage of the generated databases; and 5) In the online phase, before fusing fingerprinting and PDR solutions, the reliable relative PDR displacement and gyro heading change are proposed as additional criteria to boost the selection of the best nearest RPs while walking in straight portions. Detecting these portions between successive ANs was also leveraged to calibrate step length, control gyro heading drift, and filter PDR outliers. Multi-sensor integration of the measurements of standalone user can improve the solution robustness and reliability. However, standalone user-based IPSs are prone to variability and degradation under different scenarios. This can be attributed to several reasons including: the complexity and diversity of indoor environments, the low density and lousy distribution of wireless anchor nodes, and the signal propagation problems in indoor areas. Therefore, in this research, a real-time multi-user collaborative indoor localization scheme is also proposed to improve the positioning performance of standalone user-based systems. The results of tests conducted in different scenarios lead to the conclusion that, without the need to use external resources and compared to the standalone user solution, collaborative localization can improve the localization accuracy of the collaborated nodes. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/12441