How to Discover Knowledge for FutureG: Contextual RAG and LLM Prompting for O-RAN

TL;DR

This paper tackles the challenge of extracting accurate, up-to-date knowledge from rapidly evolving O-RAN specifications for Q&A tasks. It proposes Contextual RAG, which grounds LLM outputs by retrieving document chunks using candidate-answer-aware queries, enabling more targeted and disambiguated results without model fine-tuning. Through ORAN-Bench-13K and evaluations of Llama 3.2, Qwen 2.5-7B, and Qwen 3.0-4B, Contextual RAG consistently improves accuracy over vanilla RAG and direct prompting, with notable gains when combined with Chain-of-Thought prompting. The study also analyzes latency and CO2 emissions, showing that while retrieval increases cost, Contextual RAG remains a scalable, domain-specific approach with practical benefits for 5G/6G standards interpretation and deployment.

Abstract

We present a retrieval-augmented question answering framework for 5G/6G networks, where the Open Radio Access Network (O-RAN) has become central to disaggregated, virtualized, and AI-driven wireless systems. While O-RAN enables multi-vendor interoperability and cloud-native deployments, its fast-changing specifications and interfaces pose major challenges for researchers and practitioners. Manual navigation of these complex documents is labor-intensive and error-prone, slowing system design, integration, and deployment. To address this challenge, we adopt Contextual Retrieval-Augmented Generation (Contextual RAG), a strategy in which candidate answer choices guide document retrieval and chunk-specific context to improve large language model (LLM) performance. This improvement over traditional RAG achieves more targeted and context-aware retrieval, which improves the relevance of documents passed to the LLM, particularly when the query alone lacks sufficient context for accurate grounding. Our framework is designed for dynamic domains where data evolves rapidly and models must be continuously updated or redeployed, all without requiring LLM fine-tuning. We evaluate this framework using the ORANBenchmark-13K dataset, and compare three LLMs, namely, Llama3.2, Qwen2.5-7B, and Qwen3.0-4B, across both Direct Question Answering (Direct Q&A) and Chain-of-Thought (CoT) prompting strategies. We show that Contextual RAG consistently improves accuracy over standard RAG and base prompting, while maintaining competitive runtime and CO2 emissions. These results highlight the potential of Contextual RAG to serve as a scalable and effective solution for domain-specific Q&A in ORAN and broader 5G/6G environments, enabling more accurate interpretation of evolving standards while preserving efficiency and sustainability.

Figures (6)

• Figure 1: Contextual RAG retrieval framework.
• Figure 2: Inference run times across different prompting strategies for the Llama 3.2 and Qwen 2.5 models.
• Figure 3: Inference run times across different prompting strategies for the Qwen 3.0 Model.
• Figure 4: Inference run times across retrieval methods for the Llama 3.2 model.
• Figure 5: Inference run times across retrieval methods for the Qwen 2.5 model.