Accurate positioning in underground transportation systems remains a significant challenge due to the absence of Global Navigation Satellite System (GNSS) signals and the complex structure of subway environments. Many existing solutions depend on additional infrastructure such as beacons, Wi-Fi access points, or specialized sensors, which increases deployment costs and limits scalability. In this paper, we present a smartphone-based underground positioning approach that relies solely on embedded sensors, with a particular focus on barometric pressure measurements.
The proposed method is based on barometer fingerprinting, exploiting the fact that subway lines and stations exhibit distinctive and repeatable barometric pressure patterns. These patterns arise from elevation changes, tunnel gradients, station depths, ventilation effects, and train dynamics. Using off-the-shelf smartphones, we collected more than 15 hours of real-world sensor data across multiple U-Bahn lines in Munich. The dataset includes barometric pressure readings, transport mode annotations, and contextual information, allowing us to build detailed barometric pressure maps that serve as fingerprints for individual subway segments and stations. The data is filtered and downsampled to improve robustness against sensor noise and environmental variability.
To estimate the passenger’s trajectory, we integrate the barometric fingerprints into a particle filter framework. The particle filter represents multiple hypotheses of the passenger’s position along the subway network and updates them over time using barometric observations and motion constraints derived from the subway topology. This probabilistic approach enables the estimation of the subway line being used, detection of station arrivals and stops, and tracking of passenger progression along the route.
Experimental results demonstrate that barometric pressure signatures are sufficiently distinctive to discriminate between subway lines and stations, even in dense underground networks. The proposed approach achieves reliable line identification and station-level localization without relying on external infrastructure or prior user calibration. These findings highlight the potential of barometer-based fingerprinting combined with probabilistic filtering as a low-cost, energy-efficient, and scalable solution for underground positioning. The presented work contributes a practical step toward robust smartphone-based localization in subway and other GNSS-denied transportation environments.

In recent years, many location-based services and autonomous systems increasingly rely on user position information. However, in complex environments such as urban canyons, Global Navigation Satellite System (GNSS) performance is degraded by non-line-of-sight (NLOS) and multipath effects. Environmental obstacles such as buildings can block the direct line-of-sight (LOS) or lead to another reflected path (multipath) for satellite signals, introducing significant pseudorange biases and degraded measurement quality and preventing meter-level accuracy.

Previous studies have used three-dimensional (3D) maps to transform GNSS measurement errors into location-dependent features that indicate the user’s position, rather than explicitly correcting the measurements; however, these methods have notable limitations. In 2011, Groves et al. proposed a 3D mapping‑aided (3DMA) GNSS algorithm, shadow matching, which uses 3D maps to predict satellite LOS/NLOS status and compare these predictions with receiver data to improve urban positioning. Nevertheless, using carrier‑to‑noise density ratio (C/N₀)–based satellite visibility patterns as a position indicator can cause ambiguity, since many urban locations share similar sky visibility. In 2015, Hsu et al. introduced a 3D map‑based ray‑tracing method that simulates reflected paths and resultant delays for NLOS reception and estimates the most likely user position by matching simulated and measured delays at candidate locations. However, since this estimation is performed at the positioning level within a feature‑matching framework, NLOS pseudorange errors are not explicitly corrected, which limits tight integration with other technologies.

This study proposes an algorithm for correcting GNSS pseudorange errors using a ray-tracing–based positioning approach. First, 3D map data are used to simulate signal propagation and estimate measurement status (LOS, NLOS, or multipath) and the corresponding delays. Second, candidate receiver positions are sampled around the initial solution from the traditional solution, and likelihoods are evaluated by matching simulated and measured delays. Inspired by statistical aggregation used in other 3DMA GNSS methods for velocity or visibility estimation, the propagation delays and their likelihoods over all candidates are statistically summarized into a weighted NLOS delay in the local area. The contributions of this study are twofold: (1) it systematically assesses the potential of 3D map–based ray-tracing for statistically correcting GNSS pseudorange measurements at the true user position; (2) it introduces a 3D map–aided GNSS NLOS correction method that models local NLOS delay statistics and applies likelihood-based delay matching to obtain weighted pseudorange correction.

The experimental results demonstrate the effectiveness of the proposed algorithm. We use the mean absolute error (MAE) (meters) and standard deviation (STD) (meters) of pseudorange errors as key indicators of NLOS error distributions and correction performance. At the ground-truth location, the MAE and STD for NLOS satellites change from 58.9 m and 25.4 m to 14.7 m and 25.1 m, indicating strong potential at the true user position. After applying likelihood-based matching to real measured data, they become 20.7 m and 31.6 m, still showing a clear reduction in systematic bia. Across four test datasets, the correction not only reduces the mean pseudorange error but also yields a residual error distribution that is more symmetric and more tightly centered around zero.

The deployment of Smart Road infrastructures relies on a dense network of RSUs (Roadside Units) to enable advanced ITS (Intelligent Transport Systems) services to enhance road safety by means communication-based approach for smart traffic management.
These RSUs are currently equipped with communication systems and perception sensors, as well as commercial GNSS receivers to support the timing requirements of the underlying telecommunications network. Due to the limited communication range of RSUs, their mutual spacing is generally small, resulting in an inherently dense and spatially distributed GNSS receiver network. This feature naturally enables the possibility of leveraging this dense network of stations to implement advanced GNSS augmentation services, ensuring high-accuracy and integrity localization for vehicles and road users. Furthermore, a potential evolution of this approach is the possibility of adopting multi–base station GNSS augmentation techniques, such as multibaseline PPP-RTK, which are specifically designed to leverage dense reference networks to enhance positioning performance.
While multibaseline PPP-RTK algorithms are well established in the literature, their applicability and benefits within Smart Road scenarios have not yet been thoroughly investigated.
The objective of this paper is to provide a comprehensive performance evaluation of a multibaseline PPP-RTK framework when the augmentation network is formed by GNSS receivers installed on Smart Road RSUs. In particular, the study focuses on the performance analysis of Smart Road architectures supporting multi-baseline PPP-RTK service, evaluating accuracy and integrity across different network topologies and varying inter-RSU distances. Furthermore, the impact of using a subset of RSUs for integrity monitoring of the reference network supporting PVT (position, velocity, and time) estimation is analyzed. From a system engineering perspective, the study also aims to investigate trade-offs between RSU density, performance improvements and costs associated with deploying and operating infrastructure-based augmentation services for smart road applications.
The analysis is conducted entirely in a simulated way through the VIRGILIO software, developed by Radiolabs which enables realistic GNSS PVT performance assessment in multi-sensor approach, and relevant GNSS data, acquired in real operational environments or synthetically generated to simulate complex GNSS scenarios. The results demonstrate that implementing a multibaseline PPP-RTK service over a Smart Road network significantly enhances both the accuracy and integrity of the PVT estimated by onboard systems, while providing quantitative insights to achieve an optimal trade-off between network density, system performance, and deployment and maintenance costs.