부산대학교 도서관

Driven by the key principles of open interfaces, virtualization and programmability, Open RAN has emerged as a new paradigm to evolve contemporary Radio Access Networks (RANs) into a more vendor-agnostic, softwarized, and intelligent ecosystem. To this end, Software Defined RAN (SD-RAN) initiatives (e.g., O-RAN) are drafting specifications to provide the means to embrace it. However, even though O-RAN following the Open RAN paradigm specifies Service Models (SMs) to monitor and control the RAN, it does not go beyond the Quality of Service (QoS) mechanisms provided by 3GPP. Therefore, the QoS degradation that occurs mostly due to data flow’s nature at the slowest data path link (e.g., high L2 sublayers), is not addressed by contemporary O-RAN SMs. In this paper, we present a traffic control system for SD-RAN, denoted as TC-RAN, that consists of an E2 service Model (E2SM) and a RAN Function (RF) that adheres to the Open RAN principles and promotes data flows to first-class citizens in cellular networks, upgrading contemporary 5G QoS mechanism. TC-RAN introduces a 6 programmable, extendable, and customizable pipeline composed of a classifier, a policer, a queue, a scheduler, a shaper and a pacer. Additionally, TC-RAN addresses QoS degradation scenarios unsolvable through Resource Block (RB) allocation or 3GPP slicing mechanisms, unleashing the true potential for deploying extremely demanding applications and creating a green field for AI/ML cross-optimization algorithms on the road to 6G. We prototype and validate TC-RAN in a real 5G Stand Alone (SA) RAN stack using an O-RAN compatible near Real-Time Radio Intelligent Controller (nearRT-RIC), xApps, and Commercial off-the-shelf (COTS) User Equipments (UEs). The results show that intelligently composing a TC-RAN pipeline in cellular networks can considerably reduce the latency, notably enhancing the Quality of Experience (QoE) in a real multiplayer online game.