Traffic Matrix Estimation based on Denoising Diffusion Probabilistic Model

TL;DR

The powerful ability of denoising diffusion probabilistic models (DDPMs) on distribution learning is leveraged, and for the first time, DDPMs are adopted to address the traffic matrix estimation problem.

Abstract

The traffic matrix estimation (TME) problem has been widely researched for decades of years. Recent progresses in deep generative models offer new opportunities to tackle TME problems in a more advanced way. In this paper, we leverage the powerful ability of denoising diffusion probabilistic models (DDPMs) on distribution learning, and for the first time adopt DDPM to address the TME problem. To ensure a good performance of DDPM on learning the distributions of TMs, we design a preprocessing module to reduce the dimensions of TMs while keeping the data variety of each OD flow. To improve the estimation accuracy, we parameterize the noise factors in DDPM and transform the TME problem into a gradient-descent optimization problem. Finally, we compared our method with the state-of-the-art TME methods using two real-world TM datasets, the experimental results strongly demonstrate the superiority of our method on both TM synthesis and TM estimation.

Figures (7)

• Figure 1: Graphical models for diffusion in DDPM.
• Figure 2: Overview of the proposed DDPM-based TME method.
• Figure 3: The t-SNE and PCA plots of synthetic TMs in the Abilene dataset: the red dots denote the true TM data, and the blue dots denote the synthetic TM data.
• Figure 4: The t-SNE and PCA plots of synthetic TMs in the GÉANT dataset: the red dots denote the true TM data, and the blue dots denote the synthetic TM data.
• Figure 5: Temporal estimation errors.