A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Remote Sensing and Geographic Information Systems, School of Engineering and Technology

Thesis (Ph.D.) - Asian Institute of Technology, 2019

Global Navigation Satellite System (GNSS) is an important tool for daily life in modern society. However, in an urban area where there is the most demand for positioning services, the performance of GNSS is significantly reduced by the 'Urban canyon' effect. Tall buildings and narrow streets in a city create a canyon-like environment that prevents satellite signals to reach a receiver. In general, the positioning performance of GNSS receivers depends on the number of visible satellites. Therefore, when evaluating the GNSS availability and performance in urban areas, local urban canyon geometry should be taken into account. Conventionally, 3D city models were used to provide urban canyon geometry. However, 3D city models are not widely available and are costly to create. On the other hand, Google Street View imagery is available in most cities. This study proposes a simulation of GNSS visibility using urban canyon geometry extracted from Google Street View imagery. The first step of the simulation is to acquire multiple Street View images and then combine them together to create a sky view image that looks similar to a photo taken by a fisheye camera pointing to the sky. The next step is to extract urban canyon geometry in the sky view image, using a simple image processing algorithm. The result is called a sky mask image, a binary image where white represents the sky area, and black represents the non-sky area. The positions of GNSS satellites are calculated by using the SGP4 orbital model and the TLE orbital parameters of the satellites. Finally, the positions of the satellites are tested with a sky mask image to determine their visibility. To validate the simulation, two experiments were conducted in Bangkok and Tokyo. The results showed that the simulation can predict the visibility of GNSS satellites with relatively decent accuracy. The results of a 100-square-kilometer simulation in central Bangkok and a web application of the simulation are also presented.