Temporal Characterization and Prediction of VR Traffic: A Network Slicing Use Case

Over the past few years, concept of Virtual Reality (VR) has attracted increasing interest thanks to its extensive industrial and commercial applications. Currently, 3D models virtual scenes are generally stored in VR visor itself, which operates as a standalone device. However, applications that entail multi-party interactions will likely require scene be processed by an external server then streamed visors. stringent Quality Service (QoS) constraints imposed VR's interactive nature Network Slicing (NS) solutions, for profiling traffic generated application is crucial. To this end, we collected more than 4 hours traces real setup analyzed their temporal correlation, focusing on CBR encoding mode, should generate predictable streams. From data, distilled two prediction future frame size, can instrumental design dynamic resource allocation algorithms. Our results show even state-of-the-art H.264 mode may have significant size fluctuations, impacting NS optimization. We exploited dynamically determine requirements scenario, providing required QoS while minimizing usage.