Accurate Network Traffic Matrix Prediction via LEAD: an LLM-Enhanced Adapter-Based Conditional Diffusion Model

TL;DR

This work tackles accurate network Traffic Matrix forecasting under bursty, non-stationary dynamics with uncertainty considerations. It introduces LEAD, a framework that treats TM forecasting as a semantic-guided generative task by translating TMs into RGB images, leveraging a frozen LLM with trainable adapters for high-level temporal reasoning, and guiding a conditional diffusion model with dual global and sequential conditioning. The approach yields substantial accuracy gains on Abilene and GEANT, with RMSE reductions of up to 45.2% over strong baselines and robust long-horizon performance, while providing uncertainty-aware generation through diffusion sampling. The findings highlight the practicality of integrating foundation-model reasoning into edge-network predictive tasks, enabling proactive traffic engineering and SLA assurance in AI-native environments.

Abstract

Driven by the evolution toward 6G and AI-native edge intelligence, network operations increasingly require predictive and risk-aware adaptation under stringent computation and latency constraints. Network Traffic Matrix (TM), which characterizes flow volumes between nodes, is a fundamental signal for proactive traffic engineering. However, accurate TM forecasting remains challenging due to the stochastic, non-linear, and bursty nature of network dynamics. Existing discriminative models often suffer from over-smoothing and provide limited uncertainty awareness, leading to poor fidelity under extreme bursts. To address these limitations, we propose LEAD, a Large Language Model (LLM)-Enhanced Adapter-based conditional Diffusion model. First, LEAD adopts a "Traffic-to-Image" paradigm to transform traffic matrices into RGB images, enabling global dependency modeling via vision backbones. Then, we design a "Frozen LLM with Trainable Adapter" model, which efficiently captures temporal semantics with limited computational cost. Moreover, we propose a Dual-Conditioning Strategy to precisely guide a diffusion model to generate complex, dynamic network traffic matrices. Experiments on the Abilene and GEANT datasets demonstrate that LEAD outperforms all baselines. On the Abilene dataset, LEAD attains a remarkable 45.2% reduction in RMSE against the best baseline, with the error margin rising only marginally from 0.1098 at one-step to 0.1134 at 20-step predictions. Meanwhile, on the GEANT dataset, LEAD achieves a 0.0258 RMSE at 20-step prediction horizon which is 27.3% lower than the best baseline.

Figures (5)

• Figure 1: The Overall Structure of LEAD
• Figure 2: Vision Encoder Structure in Detail
• Figure 3: Adapter Structure in Detail
• Figure 4: Conditional U-Net Structure in Detail
• Figure 5: RMSE Comparison between LEAD and STGNNs & Multi-Scale Methods