Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

REAL: Reinforcement Learning-Enabled xApps for Experimental Closed-Loop Optimization in O-RAN with OSC RIC and srsRAN

Ryan Barker, Alireza Ebrahimi Dorcheh, Tolunay Seyfi, Fatemeh Afghah

TL;DR

The paper tackles real-time AI-driven network slicing in disaggregated O-RAN by building an end-to-end, open-source testbed that couples the OSC near-RT RIC with srsRAN and uses online PPO-based reinforcement learning to allocate PRBs across URLLC, eMBB, and mMTC slices. It integrates GNU Radio-based channel emulation to create urban mobility scenarios and enables closed-loop control with KPI feedback every 500 ms, demonstrating improved QoS across slices for up to 12 UEs while revealing practical scalability and computational challenges. Key contributions include the full-stack OSC RIC–srsRAN integration, online closed-loop training, a slice-aware reward design with per-slice objectives, and actionable insights on open-source constraints and potential scalability paths. The work provides a baseline for real-time RL-driven slicing in a lightweight 5G framework and highlights directions like hybrid online/offline training and multi-agent RL to scale to larger deployments.

Abstract

Open Radio Access Network (O-RAN) offers an open, programmable architecture for next-generation wireless networks, enabling advanced control through AI-based applications on the near-Real-Time RAN Intelligent Controller (near-RT RIC). However, fully integrated, real-time demonstrations of closed-loop optimization in O-RAN remain scarce. In this paper, we present a complete framework that combines the O-RAN Software Community RIC (OSC RIC) with srsRAN for near-real-time network slicing using Reinforcement Learning (RL). Our system orchestrates resources across diverse slice types (eMBB, URLLC, mMTC) for up to 12 UEs. We incorporate GNU Radio blocks for channel modeling, including Free-Space Path Loss (FSPL), single-tap multipath, AWGN, and Doppler effects, to emulate an urban mobility scenario. Experimental results show that our RL-based xApps dynamically adapt resource allocation and maintain QoS under varying traffic demands, highlighting both the feasibility and challenges of end-to-end AI-driven optimization in a lightweight O-RAN testbed. Our findings establish a baseline for real-time RL-based slicing in a disaggregated 5G framework and underscore the need for further enhancements to support fully simulated PHY digital twins without reliance on commercial software.

REAL: Reinforcement Learning-Enabled xApps for Experimental Closed-Loop Optimization in O-RAN with OSC RIC and srsRAN

TL;DR

The paper tackles real-time AI-driven network slicing in disaggregated O-RAN by building an end-to-end, open-source testbed that couples the OSC near-RT RIC with srsRAN and uses online PPO-based reinforcement learning to allocate PRBs across URLLC, eMBB, and mMTC slices. It integrates GNU Radio-based channel emulation to create urban mobility scenarios and enables closed-loop control with KPI feedback every 500 ms, demonstrating improved QoS across slices for up to 12 UEs while revealing practical scalability and computational challenges. Key contributions include the full-stack OSC RIC–srsRAN integration, online closed-loop training, a slice-aware reward design with per-slice objectives, and actionable insights on open-source constraints and potential scalability paths. The work provides a baseline for real-time RL-driven slicing in a lightweight 5G framework and highlights directions like hybrid online/offline training and multi-agent RL to scale to larger deployments.

Abstract

Open Radio Access Network (O-RAN) offers an open, programmable architecture for next-generation wireless networks, enabling advanced control through AI-based applications on the near-Real-Time RAN Intelligent Controller (near-RT RIC). However, fully integrated, real-time demonstrations of closed-loop optimization in O-RAN remain scarce. In this paper, we present a complete framework that combines the O-RAN Software Community RIC (OSC RIC) with srsRAN for near-real-time network slicing using Reinforcement Learning (RL). Our system orchestrates resources across diverse slice types (eMBB, URLLC, mMTC) for up to 12 UEs. We incorporate GNU Radio blocks for channel modeling, including Free-Space Path Loss (FSPL), single-tap multipath, AWGN, and Doppler effects, to emulate an urban mobility scenario. Experimental results show that our RL-based xApps dynamically adapt resource allocation and maintain QoS under varying traffic demands, highlighting both the feasibility and challenges of end-to-end AI-driven optimization in a lightweight O-RAN testbed. Our findings establish a baseline for real-time RL-based slicing in a disaggregated 5G framework and underscore the need for further enhancements to support fully simulated PHY digital twins without reliance on commercial software.

Paper Structure

This paper contains 21 sections, 4 equations, 2 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction and Related Work
  2. RELATED WORK
  3. System Model
  4. Architecture Overview
  5. Channel Assumptions and Modeling
  6. Methodology
  7. Reinforcement Learning Framework for RAN Slicing Optimization
  8. State and Action Spaces
  9. Slice-Specific Reward Functions
  10. URLLC Reward ($r_{\text{URLLC}}$).
  11. eMBB Reward ($r_{\text{eMBB}}$).
  12. mMTC Reward ($r_{\text{mMTC}}$).
  13. Clipping and Weights.
  14. Training Procedure
  15. Resource Management Constraints: Scalability Limitations and Open-Source Commitment
  16. ...and 6 more sections

Figures (2)

  • Figure 1: Overall system architecture. The near-RT RIC controls the srsRAN gNB via the E2 interface, while GNU Radio injects channel impairments.
  • Figure 2: Comparison of network slice CDFs.