Modelling of crowdsourced Wi-Fi fingerprint data
Kannan, Siva Sankar (2024)
2024
Master's Programme in Automation Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2024-06-04
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202405306532
https://urn.fi/URN:NBN:fi:tuni-202405306532
Tiivistelmä
The lack of and/or the unreliability of GPS signals indoors poses unique challenges with accurate indoor navigation. This thesis proposes an idea that aims to addresses these challenges by leveraging Wi-Fi fingerprinting to augment Pedestrian Dead Reckoning (PDR) based Inertial Navigation Systems (INS).
Wi-Fi fingerprinting involves the collection of Wi-Fi signal strengths from multiple access points, which are then used to model the relationship between fingerprint dissimilarity and real-world distances. Wi-Fi fingerprint data can be modelled through crowdsourced Wi-Fi fingerprint data. This model is crucial for enhancing indoor navigation accuracy where GPS data is unavailable. The research introduces a sophisticated approach using Weighted Least Squares regression with linear scaling weights to refine the estimation process. The Wi-Fi fingerprint model is used to filter out unreliable PDR data, considerably improving the location estimation accuracy. It employs a dual-model approach that allows utilisation of known reference points such as GPS fixes when available or Bluetooth beacons as indoor landmarks to further enhance the reliability of the navigation system.
A weighted algorithm prioritizing data points based on their estimated reliability, effectively reducing the influence of poor-quality data on the overall system performance is used. This method shows a marked improvement in positioning accuracy, thus demonstrating the feasibility and effectiveness of integrating Wi-Fi fingerprinting with traditional inertial navigation methods.
The findings showcase the potential of using Wi-Fi fingerprint modelling as a powerful augmenting technology for PDR-based INS (Inertial Navigation System), offering improvements over existing methods, particularly in complex indoor environments. The research also lays the groundwork for future advancements in indoor navigation technologies, opening avenues for more reliable and accurate indoor positioning solutions that can operate without GPS.
Wi-Fi fingerprinting involves the collection of Wi-Fi signal strengths from multiple access points, which are then used to model the relationship between fingerprint dissimilarity and real-world distances. Wi-Fi fingerprint data can be modelled through crowdsourced Wi-Fi fingerprint data. This model is crucial for enhancing indoor navigation accuracy where GPS data is unavailable. The research introduces a sophisticated approach using Weighted Least Squares regression with linear scaling weights to refine the estimation process. The Wi-Fi fingerprint model is used to filter out unreliable PDR data, considerably improving the location estimation accuracy. It employs a dual-model approach that allows utilisation of known reference points such as GPS fixes when available or Bluetooth beacons as indoor landmarks to further enhance the reliability of the navigation system.
A weighted algorithm prioritizing data points based on their estimated reliability, effectively reducing the influence of poor-quality data on the overall system performance is used. This method shows a marked improvement in positioning accuracy, thus demonstrating the feasibility and effectiveness of integrating Wi-Fi fingerprinting with traditional inertial navigation methods.
The findings showcase the potential of using Wi-Fi fingerprint modelling as a powerful augmenting technology for PDR-based INS (Inertial Navigation System), offering improvements over existing methods, particularly in complex indoor environments. The research also lays the groundwork for future advancements in indoor navigation technologies, opening avenues for more reliable and accurate indoor positioning solutions that can operate without GPS.