Chain-of-Thought Reasoning Enhances In-Context Learning for LLM-Based Mobile Traffic Prediction

2026-05-09

AI Summary

This study addresses numerical instability and limited temporal reasoning capability in short-term mobile traffic forecasting for 5G/6G networks by proposing a large language model (LLM)-based prediction framework grounded in a Planned Chain-of-Thought (PCoT) mechanism. The approach constructs structured CoT exemplars offline and retrieves, in near real time, the most similar historical samples based on current traffic patterns and short-term dynamics to enable contextual learning. A novel “plan–lecture–rationale” PCoT pipeline is introduced to significantly enhance prediction accuracy and robustness. Experiments on real-world 5G traffic data demonstrate that the proposed method outperforms both 2-shot in-context learning LLMs and classical baselines, achieving relative improvements of up to 14.88%, 15.03%, and 22.41% in MAE, RMSE, and R², respectively; further optimization of the number of contextual examples yields additional gains of 4.58%, 5.70%, and 4.85%.

Abstract

Accurate short-term mobile traffic prediction is important for proactive resource allocation and low-latency network management in fifth generation (5G) and sixth generation (6G). While large language models (LLMs) can perform in-context learning (ICL) without task-specific retraining, naive ICL prompting may suffer from numerical instability and limited temporal reasoning when traffic dynamics fluctuate rapidly. In this paper, we propose a chain-of-thought (CoT)-enabled LLM-based mobile traffic prediction framework that operates in two phases: (i) an offline phase that constructs structured CoT demonstrations by generating rationales via a plan-based CoT (PCoT) pipeline (lecture, plan, and rationale), and (ii) an online phase that performs close to real-time prediction by retrieving the most relevant demonstrations using a similarity policy that considers both the historical throughput pattern and its short-term changes. We evaluate the proposed framework using a real-world 5G measurement dataset that includes both driving and static scenarios across diverse applications. Our numerical results reveal that the proposed 2-shot CoT-LLM can improve mean absolute error (MAE), root mean square error (RMSE) and R2-score by up to 14.88%, 15.03%, and 22.41%, respectively, compared to the 2-shot ICL-LLM and classical baselines. Furthermore, by optimizing the number of in-context examples, we achieve additional improvements of 4.58%, 5.70%, and 4.85% in MAE, RMSE, and R2-score, respectively.

Problem

• mobile traffic prediction
• in-context learning
• chain-of-thought reasoning
• LLM
• temporal reasoning

Innovation

• Chain-of-Thought
• In-Context Learning
• Mobile Traffic Prediction
• Large Language Models
• 5G/6G Networks