LLM-Based Intent Processing and Network Optimization Using Attention-Based Hierarchical Reinforcement Learning

TL;DR

Problem: the paper addresses scalable, QoS-aware orchestration of network resources in Open RAN given operator intents. Approach: a three-stage pipeline combining (1) LLM-based intent processing with few-shot prompts, (2) Transformer-based predictive validation against future traffic via thresholds $Th_p$ and $Th_t$, and (3) attention-guided hierarchical reinforcement learning (h-DQN) to orchestrate multiple DRL-based applications. Contributions: (i) lightweight, prompt-driven intent extraction, (ii) QoS-drift-aware validation framework using $D_{QoS}$ and $A_{QoS}$, and (iii) an attention mechanism that reduces the action space to $A_s^{att}$ and coordinates $App1$–$App5$ via meta-controller/controller. Findings: simulations show a throughput gain of $12.02\%$, delay reduction of $26.5\%$, and energy-efficiency gain of $17.1\%$ over an HRL baseline, with even larger gains versus a DRL baseline (e.g., $17.25\%$, $48.6\%$, $39.3\%$). Significance: demonstrates a practical, scalable method for end-to-end intent-to-action automation in O-RAN that improves multiple KPIs while mitigating QoS drift.

Abstract

Intent-based network automation is a promising tool to enable easier network management however certain challenges need to be effectively addressed. These are: 1) processing intents, i.e., identification of logic and necessary parameters to fulfill an intent, 2) validating an intent to align it with current network status, and 3) satisfying intents via network optimizing functions like xApps and rApps in O-RAN. This paper addresses these points via a three-fold strategy to introduce intent-based automation for O-RAN. First, intents are processed via a lightweight Large Language Model (LLM). Secondly, once an intent is processed, it is validated against future incoming traffic volume profiles (high or low). Finally, a series of network optimization applications (rApps and xApps) have been developed. With their machine learning-based functionalities, they can improve certain key performance indicators such as throughput, delay, and energy efficiency. In this final stage, using an attention-based hierarchical reinforcement learning algorithm, these applications are optimally initiated to satisfy the intent of an operator. Our simulations show that the proposed method can achieve at least 12% increase in throughput, 17.1% increase in energy efficiency, and 26.5% decrease in network delay compared to the baseline algorithms.

Figures (4)

• Figure 1: Three-step methodology for intent processing, validation, and performance optimization.
• Figure 2: Impact of intent validation on energy efficiency.
• Figure 3: Impacts of operator intents on throughput.
• Figure 4: Performance analysis of the proposed method: (a) network delay, (b) throughput, and (c) energy efficiency.