Localization in the Parking Lot by Parked-Vehicle Assistance

摘要

Vehicular ad hoc networks represent a promising solution to vehicle navigation in parking lots. Many proposed architectures so far rely upon the roadside units (RSUs) around the parking lot for localization purposes. In this paper, we present a parked-vehicle assistance (PVA)-based position estimation scheme. The many PVA links we exploit significantly increase the number of communication links and range measurements used for localization. This way, we alleviate the constraint imposed by the usually limited number of RSUs. While the proposed method serves as a promising candidate to achieve accurate vehicle localization particularly in parking lots, the range measurements from the parked vehicles may suffer from severe non-line-of-sight (NLOS) conditions native to the parking lot topologies. In response, we propose a two-step algorithm to efficiently detect the NLOS links. This way, we can exploit the benefits offered by PVA, while keeping a reasonable computational load by efficiently differentiating the range measurements. Indeed, our simulation results demonstrate a significant increase in terms of the position accuracy compared with the RSU-based scheme.