Evaluation of 5G Cell Densification for Autonomous Vehicles Positioning in Urban Settings

A key operational requirement for Autonomous vehicles (AVs) is to have a highly reliable positioning at the sub-meter lane-level accuracy. However, it is well-known that current satellite- and perception-based positioning systems suffer in achieving this desired accuracy in urban settings and during rough weather conditions. This research explores the strong potentials of the soon-to-be-deployed 5G wireless technology that is capable of overcoming these limitations and provides an uninterrupted everywhere positioning with lane level accuracy. The high cell densities and large bandwidth ranges promised in 5G are anticipated to achieve ultra-reliable and ultra-low-latency communications, which will enable the detection of received signals at AVs with high time precision, thus improving the localization accuracy. This paper discusses the merits and limitations of using 5G small cells to provide lane level positioning services in urban environments. We consider a 5G-based positioning scheme employing time of arrival with the trilateration of ranges between 5G base stations based on least-squares and an AV in kinematic mode. We then evaluate the impact of cell densification on achieving the desired accuracy level using the considered positioning scheme. A professional 5G simulator was used to assess the positioning accuracy of a vehicle moving at an average speed of 35 km/h in a kinematic road test involving different 5G base-stations densities on a trajectory in downtown Manhattan, NY. Results show that an inter-cell spacing of 160 m can achieve sub-meter positioning accuracy for AVs in typical dense urban settings.