Vehicle-to-Everything (V2X) Datasets for Machine Learning-based Predictive Quality of Service

Aug 5, 2026

Presents two machine learning datasets for Predictive Quality of Service (PQoS) in Vehicle-to-Everything (V2X), built from Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) radio channel measurements. Combines both datasets to study V2X connectivity across its full complexity. Details measurement-campaign methodologies for V2V and V2I, and provides usage examples. Demonstrates approximate Bayesian methods for uncertainty-aware Quality of Service and Channel State Information forecasting, then outlines future exploratory research. Datasets are hosted on IEEE DataPort; experiment code is available via CodeOcean.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Introduction

Predictive QoS and proactive network management

Motivation for V2X predictive quality of service

Dataset need and proposed contributions

Distinct V2V and V2I characteristics

V2V measurement focus (mmWave, 60 GHz)

V2I measurement focus (sub-6 GHz, KPI-oriented)

Vehicle-to-Everything (V2X) Datasets for Machine Learning-based Predictive Quality of Service Marco Skocaj∗ , Nicola Di Cicco†, Tommaso Zugno‡, Mate Boban‡, Jiri Blumenstein§ , Ales Prokes§ , Tomas Mikulasek§ , Josef Vychodil§ , Konstantin Mikhaylov¶ , Massimo Tornatore† and Vittorio Degli-Esposti∗ ∗ DEI, University of Bologna, & WiLab, CNIT, Italy † DEIB, Politecnico di Milano, Italy ‡ Huawei Technologies § Brno University of Technology, Czech Republic ¶ Centre for Wireless Communications (CWC), University of Oulu (UO), Finland corresponding [authors:](authors: marco.skocaj@unibo.it)[ marco.skocaj@unibo.it](authors: marco.skocaj@unibo.it), [nicola.dicicco@polimi.it](nicola.dicicco@polimi.it) Abstract—We present two datasets for Machine Learning (ML)-based Predictive Quality of Service (PQoS) comprising Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) radio channel measurements. As V2V and V2I are both indispensable elements for providing connectivity in Intelligent Transport Systems (ITS), we argue that a combination of the two datasets enables the study of Vehicle-to-Everything (V2X) connectivity in its entire complexity. We describe in detail our methodologies for performing V2V and V2I measurement campaigns, and we provide illustrative examples on the use of the collected data. Specifically, we showcase the application of approximate Bayesian Methods using the two presented datasets to portray illustrative use cases of uncertainty-aware Quality of Service and Channel State Information forecasting. Finally, we discuss novel exploratory research direction building upon our work. The V2I and V2V datasets are available on IEEE Dataport1 , and the code utilized in our numerical experiments is publicly accessible via CodeOcean2. Index Terms—predictive QoS, V2X, V2V, V2I, machine learning, deep learning. I. INTRODUCTION One of the core objectives of next-generation Beyond 5G (B5G) and 6G wireless systems is the minimization of human intervention in network management. On the one hand, network operators need new and more cost-effective solutions to match the increased management complexity. On the other hand, new services like autonomous driving set stringent requirements on reliability and guaranteed Quality of Service (QoS) . As such, integrating autonomous capabilities and proactive decision making in network management constitutes a timely research challenge. Predictive Quality of Service (PQoS) was introduced as a real-time mechanism to provide autonomous systems with advance notifications about upcoming QoS changes [3] . In contrast to reactive strategies, PQoS allows for proactive decision-making and ensures agile adaptation and continuity of service following a predict-adaptinform closed loop principle. In Vehicle-to-everything (V2X) applications (e.g., trajectory prediction, high-density platooning, teleoperated driving, etc. [3]) PQoS was introduced to accommodate various configurations (such as automation level, inter-vehicle gap, etc.) and allows for configuration M. Skocaj and N. Di Cicco are co-first authors of this paper. 1Links to IEEE DataPort datasets: [1] [https://dx.doi.org/10.21227/r1wm-6a24](https://dx.doi.org/10.21227/r1wm-6a24) [2] [https://dx.doi.org/10.21227/cfvn-hp41](https://dx.doi.org/10.21227/cfvn-hp41) 2Link to CodeOcean: [https://codeocean.com/capsule/3205606/tree](https://codeocean.com/capsule/3205606/tree) adjustments in response to changes in QoS. Prompt adjustment may favor service continuity, comfort, and safety. Besides that, predictive knowledge of future network conditions might trigger a series of network procedures aimed at ameliorating channel conditions and QoS (e.g., early handover decision or Up Link (UL)/Down Link (DL) power control) . Recently, two main factors contributed to the rise of PQoSand autonomous adaptivity in network management: technological breakthroughs in the fields of artificial intelligence (AI) and machine learning (ML

FAQs

What do the two proposed datasets cover in V2X predictive QoS?

They cover Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) radio channel measurements, enabling machine learning-based Predictive Quality of Service (PQoS) studies for V2X connectivity.

Why is combining the V2V and V2I datasets important?

V2V and V2I are both essential for Intelligent Transport Systems connectivity; using both datasets supports studying V2X connectivity in its full complexity.

How are the datasets used to demonstrate predictive performance?

The work showcases approximate Bayesian methods using the collected data for uncertainty-aware Quality of Service and Channel State Information forecasting.

Aug 5, 2026

Upload by Lucas Martin