# QoS Evaluation and Prediction for C-V2X Communication in Commercially-Deployed LTE and Mobile Edge Networks

arXiv: https://arxiv.org/abs/2002.07883 · HTML: https://mlsanity.com/paper/2002.07883/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Feb 2020.

## Built on by (most cited first, 6 of 6)

- [Berlin V2X: A Machine Learning Dataset from Multiple Vehicles and Radio Access Technologies](https://mlsanity.com/paper/2212.10343/) (2022)
- [Machine Learning for QoS Prediction in Vehicular Communication: Challenges and Solution Approaches](https://mlsanity.com/paper/2302.11966/) (2023)
- [QoS prediction in radio vehicular environments via prior user information](https://mlsanity.com/paper/2402.17689/) (2024)
- [A Comprehensive Study on Artificial Intelligence Algorithms to Implement Safety Using Communication Technologies](https://mlsanity.com/paper/2205.08404/) (2022)
- [PRATA: A Framework to Enable Predictive QoS in Vehicular Networks via Artificial Intelligence](https://mlsanity.com/paper/2507.14211/) (2025)
- [A Survey on Deep Multi-Task Learning in Connected Autonomous Vehicles](https://mlsanity.com/paper/2508.00917/) (2025)

## Abstract (arXiv)

Cellular vehicle-to-everything (C-V2X) communication is a key enabler for future cooperative automated driving and safety-related applications. The requirements they demand in terms of Quality of Service (QoS) performance vary according to the use case. For instance, Day-1 applications such as Emergency Brake Light warning have less strict requirements than remote driving. In this paper, we seek to answer two questions: Are current LTE networks ready to support Day-1 applications at all times? And, can underperforming situations be reliably predicted based on GPS and network-related information? To address these questions, we first implement a system that collects positioning data and LTE key performance indicators (KPIs) with a higher time resolution than commercial off-the-shelf LTE modems, while simultaneously measuring the end-to-end (E2E) delay of an LTE network. We then use this system to assess the readiness of multiple mobile network operators (MNOs) and a live Mobile Edge Computing (MEC) deployment in an urban scenario. For evaluating whether an adaptable operation is possible in adverse circumstances, e.g., by performing hybrid networking or graceful degradation, we finally use Machine Learning to generate a client-based QoS predictor and forecast the achievable QoS levels.