Estimating quality of experience in multicast point cloud streaming over 5G networks

Fiorina, Guillermo;Wildbaum, Martin;González, Claudia Carballo;Pupo, Ernesto Fontes;Floris, Alessandro;Porcu, Simone;Murroni, Maurizio;Sotelo, Rafael

2025-01-01

---

Abstract

This paper proposes three Quality of Experience (QoE) prediction models tailored for multicast point cloud (PC) streaming, leveraging the Open Radio Access Network (O-RAN) framework for facilitating data sharing between Over-The-Top (OTT) providers and Mobile Network Operators (MNOs). The first model, called Network-only, is based on machine learning algorithms (ML) to capture the relationship between diverse 5G network parameters and the corresponding PC quality estimated using a state-of-the-art No-Reference (NR) PC quality assessment (PCQA) metric. The other two models (Network+PCCI and Network+Distortion) are trained with the same network features plus additional PC data, namely, the proposed point cloud complexity index (PCCI) and the distortion type, respectively. The PCCI is an original metric that categorizes PCs based on their inherent complexity characteristics. The achieved results demonstrate that the proposed PCCI enables the Network+PCCI QoE model to achieve a higher Pearson linear correlation coefficient (PLCC) (0.941 vs. 0.809 vs. 0.612) and a lower root mean square error (RMSE) (0.174 vs. 0.268 vs. 0.509) compared to Network+Distortion and Network-only QoE models, respectively.