A New Time Series Forecasting Approach Using Classification: Application to Field of View Prediction in 360° videos - Université de Bourgogne Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

A New Time Series Forecasting Approach Using Classification: Application to Field of View Prediction in 360° videos

Résumé

Multimedia applications based on 360° video use a remote server and require a very high bandwidth. Thus, the full transmission of the video from the server can have a negative impact on the quality of the traffic that passes through different communication networks to the end users. One of the solutions to reduce the throughput of 360° videos is to introduce the notion of field of view (FoV) which allows to transmit at a given time only a portion of the video with a higher quality. The remaining portions of the video will be transmitted with a lower quality. In this work, a lossy forecasting approach is proposed to predict the motion trajectory of users of a 360° streaming system in order to ensure better quality of experience (QoE). A new Time Series Forecasting approach using Classification (TSFC) that transforms real data into discrete data so that classification algorithms can be used to make forecasts from them. Our approach can provide online and offline learning and can guarantee live streaming where it ensures a prediction time under 6ms and remarkably well below existing approaches.
Fichier non déposé

Dates et versions

hal-04353783 , version 1 (19-12-2023)

Identifiants

Citer

Ahmed Saadallah, Ines El-Korbi, Sidi-Mohammed Senouci, Philippe Brunet. A New Time Series Forecasting Approach Using Classification: Application to Field of View Prediction in 360° videos. 2023 IEEE 97th Vehicular Technology Conference (VTC2023-Spring), Jun 2023, Florence, Italy. pp.1-5, ⟨10.1109/VTC2023-Spring57618.2023.10199671⟩. ⟨hal-04353783⟩
23 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More