Empowering immersive XR experience through scalable and cost-effective EPON access networks

Abstract

Consumers, businesses, and government organizations are expected to increasingly adopt immersive services, which are accessed in real-time using eXtended Reality (XR) devices. The 3rd Generation Partnership Project (3GPP) specifies the XR traffic model using truncated Gaussian distribution. We propose to analyze and determine optimal operating conditions when this traffic model is used in an Ethernet passive optical network (EPON) system with the “interleaved polling with adaptive cycle time (IPACT)” dynamic bandwidth allocation (DBA) scheme. This traffic demands high bandwidth, ultra-low delay, adaptive responsiveness, and precise synchronization with minimal jitter. EPON systems address these requirements effectively by eliminating bottlenecks in broadband access networks and offering cost-effective and efficient “last mile” connectivity between core and high-capacity local or home networks. This work evaluates the performance for XR traffic under the IPACT DBA scheme, focusing on delay and jitter. We aim to assess the impact of XR traffic on existing DBA mechanisms and determine the optimal frame (r<inf>f</inf>) and data rate (r<inf>d</inf>) conditions to enhance XR performance in PON systems. We analyze the key performance metrics, including end-to-end mean packet delay (d<inf>e2e</inf>), normalized jitter (J<inf>n</inf>), average grant size (G<inf>size</inf>), grant utilization (GU), average cycle time (C<inf>time</inf>), and bandwidth utilization (BWU). This paper evaluates two grant allocation policies, limited and gated services, each tailored to optimize XR traffic within the IPACT framework. Our proposed XR traffic model substantially reduces d<inf>e2e</inf> and enhances BWU, improving support for immersive applications. © 2025 Elsevier B.V., All rights reserved.