QoS Management for XR Traffic in 5G NR: A Multi-Layer System View and End-to-End Evaluation

Abstract

Extended reality (XR) applications are gaining a lot of momentum in the industry and consumer market. XR applications are characterized by multiple simultaneous data flows with different Quality-of-Service (QoS) requirements, imposing the need for advanced QoS management in 5G networks. Compared to 4G, the 5G NR technology has been designed with enhanced QoS support, by introducing the concept of QoS flows and an additional layer for QoS handling. However, recent studies show that additional QoS enhancements are needed in 5G to support the massive adoption of XR services. 3GPP envisions to include some enhancements for XR traffic in 5G-Advanced. In this paper, we review the current 5G QoS framework and study potential design improvements to fulfill XR QoS requirements in future cellular networks. We study QoS control procedures for XR traffics at different layers of the 5G NR protocol stack, including the proper configuration of radio access and core networks. In particular, we analyze various architectures for handling the peculiarities of XR, such as time-constraint and multi-flow applications. Also, we present XR traffic adaptation mechanisms at the application layer, to properly adapt XR traffic statistics to the actual 5G radio environment, and QoS schedulers for the MAC layer, leading to cross-layer performance optimization between XR application and 5G radio access network. Finally, we provide a validation study and end-to-end evaluation of QoS control mechanisms by simulating realistic mixed traffic scenarios in a 5G NR system-level simulator based on ns-3.