Mattia Brambilla

The advancement of fifth generation (5G) cellular networks offers significant opportunities for positioning services, yet the performance of these systems is currently hindered by the clock offset found in commercial networks among 5G base stations, here referred to as inter-cell clock bias (ICCB). These biases introduce systematic errors in time of arrival (TOA) measurements, degrading localization accuracy. This paper addresses this challenge by utilizing a stationary reference receiver to estimate and correct the ICCBs. By aligning the timing references of multiple base stations, our single-difference correction method enables accurate standalone positioning using opportunistically received synchronization signal block (SSB) data. The proposed method was validated through three independent measurement campaigns at the same rural location on broadcast signals from a commercial 5G network. The results demonstrate the effectiveness of the ICCB correction, showing a 73%–84% reduction in positioning error compared to uncompensated methods. Applying a smoothing filter to the ICCB corrections further improves the mean accuracy by about 18%–19%, achieving a mean error between 12.7m and 20.6m across the three tests. While this performance is constrained by the limited bandwidth of the current commercial 5G deployment at 3.68 GHz, our findings confirm the feasibility of leveraging public 5G signals for location-based services.

Accurate and reliable vehicle positioning is a prerequisite for the development of cooperative intelligent transportation systems (C-ITSs), especially for automated driving applications and in scenarios where global navigation satellite systems (GNSSs) fail to meet the requirements of safety-critical services. This article presents an experimental assessment of a roadside infrastructure-based localization system that exploits a wideband wireless technology to estimate a vehicle’s position and communicates this information in real time to the vehicle via a vehicle-to-everything (V2X) link. We highlight the need to use advanced tracking methods to cope with heterogeneous vehicle dynamics. We validate the proposed system experimentally in both racetrack and highway tunnel scenarios using an ultra wideband (UWB) positioning network, showcasing the system’s feasibility through on-field implementations. The findings reveal that the wideband radio technology obtains positioning accuracy comparable to GNSS real-time kinematic with a mismatch lower than 30 cm in 95% of the cases. This capability provides a seamless high-accuracy positioning service and successfully unlocks autonomous navigation in tunnels. This article underscores the key role of the roadside infrastructure in enhancing autonomous vehicle navigation by integrating a precise positioning technology as UWB with V2X communication, thereby meeting the stringent requirements of safety-critical C-ITS applications.

This paper presents the TRUSTroke Federated Learning (FL) Platform, supporting distributed, secure, and trustworthy training of machine learning (ML) models for collaborative healthcare applications without sharing sensitive data. The platform enables configuration, monitoring, and management of FL processes over the public Internet via a reliable and secure transport protocol based on Message Queuing Telemetry Transport (MQTT). This work details production-ready PaaS capabilities supported by an MQTT-based communication layer and a DMZ security architecture that isolates clinical data from external access, and it presents a comprehensive evaluation of network and model security that demonstrates resilience against common attacks while confirming scalability and low communication overhead. The platform is deployed and tested with geographically distributed clinical clients, supervised by a Parameter Server (PS) hosted at CERN’s computing center, and it is customized for stroke management with harmonized patient data integration. The discussion addresses practical deployment challenges and real-world application using a publicly available stroke dataset, emphasizing the importance of rigorous security assessment for FL platforms in digital healthcare contexts. Experimental validation demonstrates high model performance on stroke data (F1-score and AUPRC close to centralized training), low communication overhead (17-26 %), stable scalability across heterogeneous clinical nodes, and strong resilience to poisoning and free-rider attacks, confirming the platform’s reliability for secure digital healthcare deployments.

Internet of Things (IoT) wireless technologies serve as key enabler for location-based services in emerging applications such as autonomous robotics, industrial automation, augmented and virtual reality. Wide-Band technologies, including Ultra Wide-Band (UWB) and 5G-advanced millimeter-waves, are the preferred solutions in these contexts for their high potentials in precise positioning. A main challenge is the mitigation of radio propagation effects that arise in complex environments, such as in industrial facilities, where frequent blockage events limit the accuracy and integrity of localization services. This article tackles the problem focusing on precise indoor navigation in industrial environments with dense and dynamic blockage conditions. Our proposal relies on an innovative particle filtering technique, based on the Stein Variational Adaptive Importance Sampling (SVA-IS), to improve the sampled representation of the location posterior distribution by integrating prior information on the intermittent visibility-blockage dynamics. We assess the proposed solution through indoor experiments conducted in industrial scenarios using UWB devices. Our results show significant improvements with respect to state of the art filters in terms of both accuracy and robustness of the location tracking.

This paper studies the positioning capabilities offered by currently operational fifth-generation (5G) cellular networks operating in the n78 band. We consider four mobile operators in Milan, Italy, and compare their network readiness for positioning by passively measuring and analyzing the Time of Arrival (ToA) and the received power extracted from Synchronization Signal Blocks (SSBs) signals periodically broadcasted by the base stations. We focus on urban and rural areas, showing that the TOA accuracy significantly varies over the two types of areas and among network operators. Our operator-agnostic analysis reveals that the current state of network synchronization presents strong issues that hinder positioning services at present. Major efforts on synchronization protocols and technologies are required by telecommunication manufacturers and operators.

Road safety is one of the major concerns when considering road vehicles, with issues referring to both drivers and Vulnerable Road Users (VRUs) (e.g., pedestrians). The role of roadside infrastructures in augmenting VRU protection is gaining popularity thanks to the development of smart and connected sensing systems, which enable the detection of critical events and their forwarding to nearby Connected and Automated Vehicles (CAVs) by Vehicle-to-Everything (V2X) communications. In this context, the fifth generation (5G) of mobile radio networks is emerging as the communication enabler for advanced V2X services for CAVs. In this work we propose the design, development, and assessment on road of a 5G roadside infrastructure for VRU protection, performing an in-depth analysis of the single components and their integration into the service. More specifically, we present a Roadside Unit (RSU) capable to detect, localize, and track VRUs; a communication architecture that exploits 5G V2X connectivity for sending VRUs detections; a CAV with on-board algorithms warning the driver of the presence of the VRU and an Autonomous Emergency Braking (AEB) solution triggered by the 5G V2X architecture. We validate the efficacy of the proposed solution in a road scenario inside the campus of Politecnico di Milano.

This paper presents an experimental study on the integration of the fifth generation (5G) cellular networks and the Global Navigation Satellite System (GNSS) for positioning. Along with the use of experimental data collected from both 5G and GNSS deployed technologies, the novelty of this research includes the design of a Bayesian tracking methodology based on extended Kalman filtering for processing the 5G Time of Flights (TOFs) and jointly estimating the user position and the clock offsets of the 5G Base Stations (BSs), as well for their fusion with GNSS observations (both single and dual frequency). Data collection and experimental analyses are conducted in both static and mobile positioning scenarios. For standalone 5G positioning, we demonstrate the need of tracking the clock offsets of BSs, as they currently represent a primary impairment for precise positioning. In static conditions, we manage to reduce the positioning error from more than 100 m to approximately 5 m by jointly estimating these offsets and correct the TOFs accordingly, showcasing that the current 5G network is a possible alternative to single frequency code-only GNSS positioning. The 5G+GNSS hybrid solution is shown to guarantee a more reliable and accurate positioning as the two technologies mutually assist each other, particularly when single frequency GNSS processing is considered. On the other hand, when the GNSS receiver is able to perform dual frequency processing, location estimation does not benefit from the hybridization with 5G as GNSS is already highly accurate. Overall, the achieved results confirm the viability of 5G+GNSS integration from an experimental standpoint, showing the potentialities of the currently-deployed 5G network for positioning.

With the advent of cooperative intelligent transport systems (C-ITS) and vehicle-to-everything (V2X) communications, cooperative positioning based on V2X sharing of location information has been emerging as a promising augmentation system for conventional satellite navigation. An example is implicit cooperative positioning (ICP) which relies on Bayesian filtering for cooperative sensing of targets that are used as reference points for improving vehicle positioning. ICP methods, however, rely on pre-determined models which makes them sub-optimal in case of non-Gaussian non-linear models or complex cooperation graphs. To address these limitations, the paper proposes a decentralized-partially observable Markov decision process (Dec-POMDP) framework, paired with deep multi-agent reinforcement learning (MARL) algorithms. We introduce a novel ICP-multi-agent proximal policy optimization (MAPPO) algorithm where distributed agents (i.e., vehicles) dynamically activate/deactivate the radio links for cooperation with the neighbors to optimize the communication efficiency, still guaranteeing accurate positioning. We reproduce a realistic C-ITS scenario with CARLA simulator, where vehicles move according to real-world dynamics and communicate with each other to cooperatively sense their locations. Results show that the proposed ICP-MAPPO algorithm, with its dynamic-decentralized-execution and centralized-training schemes, outperforms state-of-the-art ICP methods by 21% in terms of positioning accuracy, and it can reduce the communication overhead by following the optimal learned policy.