Investigation of RF Fingerprinting approaches in GNSS
Gahlawat, Sarika (2020)
Degree Programme in Information Technology, MSc (Tech)
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
With the increase of emerging technologies, demand for location and positioning services is increasing in all domains of life. Currently there is an imperative need to protect and safeguard the authenticity and integrity of user data. In this thesis work Radio Frequency Fingerprinting (RFF)approaches are investigated in the context of Global Navigation Satellite System (GNSS) as the demand for authentication in location related data is increasing. The idea here is to exploit unique features that a device possesses. It is very difficult to impersonate a specific device as the device uses certain linear and non-linear components which are unique. The devices made by the exact same manufacturer are also not exactly similar because of the nature of non-linear components used in making of the devices. The purpose of this thesis work is to identify relevant features in GNSS signals at sampled domain (baseband or Intermediate Frequency (IF)) that can distinguish satellites on the sky, detect multipath, etc. The features that are addressed and studied in this work can be described as the features for identifying the source. We introduce different features and then convert the time series data signal into images by using frequency transforms. The images are then fed to machine learning algorithms. MATLAB was used for simulating this model. The images are then used as an input to the Machine Learning Algorithms (MLA)to study various device specific features. Once the features are identified, feature classification methods are applied to classify the transmitters in GNSS. The goal is to identify 3-5 top features and apply a classifier to classify the different transmitter devices. It was found out that the spectrogram and wavelet transforms applied on raw I/Q GNSS data in combination with Support Vector Machine (SVM) and Convolutional Neural Networks (CNN) gave promising results in terms of RFF. Spectrogram was the best one out of all transforms at both low and high ranges of Carrier to Noise Ratio to achieve fingerprinting in Global Navigation Satellite System.