🤖 AI Summary
To address the decoupling between IP flow control and radio resource scheduling caused by the “fat Layer 2” architecture in 5G RAN, this paper proposes the Integrated User Plane (IUP) architecture. IUP vertically integrates UPF functionality into the RAN and introduces a programmable Integrated Data Flow Control (IDFC) sublayer. It is the first work to realize both UPF integration and a programmable IDFC pipeline within the RAN, enabling real-time traffic control and joint radio resource scheduling via control-plane coordination—thereby bridging 3GPP and non-3GPP networks. Built upon the O-RAN framework and a programmable data plane, our prototype demonstrates a 50% reduction in end-to-end latency and protocol overhead. Furthermore, IUP achieves seamless interworking with heterogeneous networks and enables dual-dimensional real-time programming of user-plane traffic and radio resources.
📝 Abstract
Mobile networks evolve on a regular basis to meet the requirements of a rapidly changing application ecosystem; hence, a future-proof design is key to getting the most out of their lifecycle. In comparison to other access networks, one major issue with the 5G Radio Access Network (RAN) is that it behaves as a"fat Layer 2"entity, resulting in disparities in Internet Protocol (IP) flow traffic control and radio resource allocation. In this article, we propose an innovative design - Integrated User Plane (IUP) - that incorporates User Plane Function (UPF) functionalities into RAN, and we introduce the Integrated Data Flow Control (IDFC) sublayer with a new traffic management pipeline and various programmable rules. To understand its implications for crucial mobility user cases, a detailed analysis of how IUP interacts with Control Plane (CP) network functions is conducted. Finally, our IUP prototype shows benefits including a 50% saving in both latency and overhead, converged IUP and non-Third-Generation Partnership Project (3GPP) networks for seamless connectivity, and real-time UP programmability in both traffic control and resource allocation via the O-RAN framework.