A Solution for Improving Robustness in GNSS Positioning from Android Devices

Starting with the Nougat 7.0 version of the Android Operating System released in 2016, Google has permitted direct access, via the Android development API, to the GNSS raw measurements on Android-based smartphones. This new feature opened a new research field aimed at developing low-cost applications for satellite-based positioning systems. In particular, after the publication of a white paper from the European GNSS Agency, many authors started to analyse the quality of the raw measurements retrieved from smartphones and compare them with other types of low cost GNSS devices. The main detected issues turned out to be the low quality of the GNSS observales due to the low quality of the hardware (receiver and antenna) mounted in the smartphones. Nowadays raw GNSS measurements support is mandatory on devices that run Android 10 (API level 29) or higher, but unfortunately the support for some of the raw GNSS measurement fields (e.g. pseudorange rate, ADR, AGC) is optional and can vary based on the type of GNSS chipset installed on the device. Furthermore, not all the smartphones present on the market support double frequency or multi-constellation. In this thesis we present a solution for improving the robustness of GNSS positioning in Android devices in real time, that combines an acquisition phase performed in a dedicated Android app (thus working on the edge) and a processing phase, based on a modified version of the open source library RTKLIB, performed on dedicated servers. The processing phase performed on the smartphone is aimed at cleaning, filtering, organizing and delivering the data acquired via the GNSS raw data library. The server-side processing phase applies an improved version of the RTK library. Our version provides an additional interface for processing different format of GNSS data (in particular a new one defined for GNSS measurement from Android devices) and a method to mitigate in real-time the multipath effect on the collected data. We will focus our attention on the architecture of the proposed solution (Android endpoints and processing server), on the proposed improvement of the GNSS positioning procedure, and its implementation in our version of the RTKLIB library, and on preliminary results obtained by the resulting system.