Influence of database accuracy on two-dimensional ray-tracing-based predictions in urban microcells

Abstract

Ray-tracing-based predictions in urban microcellular environments require databases for building layouts, electrical characteristics of buildings, and base stations (locations, antennas, power, etc.). The aim of this paper is to provide help in selecting the appropriate level of accuracy required in these databases in order to achieve the best tradeoff between database costs and prediction accuracy. The effects of inaccuracies in these databases are presented and analyzed by comparing predictions and measurements. The results presented here show to what extent errors, which are due to automatic vectorization of scanned maps, could lead to erroneous predictions. Furthermore, an analysis of the influence of random errors in a building vector database was performed to quantify the prediction error as a function of the accuracy in the building vector databases. Ray-tracing prediction models implementing a reflection and diffraction phenomena were found to be sensitive to the choice of the reflection coefficient attributed to building walls. This dependence can be used to fit the measurements as the complexity of real building walls does not allow one to easily derive their electrical parameters from which a reflection coefficient could be computed. It was also found that, in general and in agreement with measurements, ray-tracing-based prediction models are not sensitive to small variations on a base-station location. Finally, the sensitivity study also lead to gained insight of the propagation phenomena involved in urban microcell environments.