Hyppää sisältöön
    • Suomeksi
    • In English
  • Suomeksi
  • In English
  • Kirjaudu
Näytä aineisto 
  •   Etusivu
  • OSUVA
  • Artikkelit
  • Näytä aineisto
  •   Etusivu
  • OSUVA
  • Artikkelit
  • Näytä aineisto
JavaScript is disabled for your browser. Some features of this site may not work without it.

Improving Precision GNSS Positioning and Navigation Accuracy on Smartphones using Machine Learning

Siemuri, Akpojoto; Selvan, Kannan; Kuusniemi, Heidi; Välisuo, Petri; Elmusrati, Mohammed S. (2021)

 
Katso/Avaa
Artikkeli (562.5Kb)
Lataukset: 

URI
https://doi.org/10.33012/2021.18004

Siemuri, Akpojoto
Selvan, Kannan
Kuusniemi, Heidi
Välisuo, Petri
Elmusrati, Mohammed S.
The Institute of Navigation
2021
doi:10.33012/2021.18004
Näytä kaikki kuvailutiedot
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe2022081755705

Kuvaus

vertaisarvioimaton
©2021 The Author. Published by ION.
Tiivistelmä
In this work, we developed a precision positioning algorithm for multi-constellation dual-frequency global navigation satellite systems (GNSS) receivers that predicts the latitude and longitude from smartphone GNSS data. Estimation for all epochs that have at least four valid GNSS observations is generated. Receivers (especially low-cost receivers) often have limited channels and computational resources, therefore, the complexity of the algorithm used in them needs to be kept low. The datasets and results in this paper are based on the data provided by Google under the session "High Precision GNSS Positioning on Smartphones Challenge" in the Institute of Navigation (ION GNSS+ 2021) conference. We began by exploring and analysing the raw GNSS data which includes the training dataset and its ground truth and the test dataset without the ground truth. This analysis gave insight into the nature and correlation of the dataset and helped shape the algorithm that was proposed for the accuracy improvement problem. The design of the algorithm was done using data science techniques to compute the average of the predictions of several devices data in the same collection (training dataset baseline coordinates and their ground truth) and then the data was used to train a few selected machine learning algorithms namely, Linear Regression (LR), Bayesian Ridge (BR) and Neural Network (NN) to predict the offset of the test data baseline coordinates from the expected ground-truth (which was not provided). A simple weighted average (SWA) which combines all the previous three ML technique was also implemented. The results showed improvement in the position accuracy with the simple weighted average (SWA) method having the best accuracy followed by Bayesian Ridge (BR), Linear Regression (LR), and then Neural Network (NN) respectively.
Kokoelmat
  • Artikkelit [3284]
https://osuva.uwasa.fi
Ota yhteyttä | Tietosuoja | Saavutettavuusseloste
 

 

Tämä kokoelma

TekijäNimekeAsiasanaYksikkö / TiedekuntaOppiaineJulkaisuaikaKokoelmat

Omat tiedot

Kirjaudu sisäänRekisteröidy
https://osuva.uwasa.fi
Ota yhteyttä | Tietosuoja | Saavutettavuusseloste