Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TUBO: A Tailored ML Framework for Reliable Network Traffic Forecasting

Zhihang Yuan, Leyang Xue, Waleed Ahsan, Mahesh K. Marina

TL;DR

Tubo tackles the challenge of reliable network DM forecasting in the presence of bursts and variable traffic patterns by introducing a tailored ML framework with a Burst Processor and uncertainty-aware Model Selector. It combines burst-aware preprocessing, multiple model types, and calibrated uncertainty via MC Dropout to select the most reliable predictor for each input, achieving up to 4x improvements in DM forecast accuracy and up to 94% burst-occurrence forecasting accuracy. The framework also demonstrates significant downstream benefits for proactive traffic engineering, delivering up to 9x throughput gains over reactive routing and 3x gains over the best existing proactive method. These results, validated on three real-world DM datasets, highlight Tubo’s potential to enhance network performance and reliability in dynamic environments, with an emphasis on practical deployment considerations and scalability.

Abstract

Traffic forecasting based network operation optimization and management offers enormous promise but also presents significant challenges from traffic forecasting perspective. While deep learning models have proven to be relatively more effective than traditional statistical methods for time series forecasting, their reliability is not satisfactory due to their inability to effectively handle unique characteristics of network traffic. In particular, the burst and complex traffic patterns makes the existing models less reliable, as each type of deep learning model has limited capability in capturing traffic patterns. To address this issue, we introduce TUBO, a novel machine learning framework custom designed for reliable network traffic forecasting. TUBO features two key components: burst processing for handling significant traffic fluctuations and model selection for adapting to varying traffic patterns using a pool of models. A standout feature of TUBO is its ability to provide deterministic predictions along with quantified uncertainty, which serves as a cue for identifying the most reliable forecasts. Evaluations on three real-world network demand matrix (DM) datasets (Abilene, GEANT, and CERNET) show that TUBO significantly outperforms existing methods on forecasting accuracy (by 4 times), and also achieves up to 94% accuracy in burst occurrence forecasting. Furthermore, we also consider traffic demand forecasting based proactive traffic engineering (TE) as a downstream use case. Our results show that compared to reactive approaches and proactive TE using the best existing DM forecasting methods, proactive TE powered by TUBO improves aggregated throughput by 9 times and 3 times, respectively.

TUBO: A Tailored ML Framework for Reliable Network Traffic Forecasting

TL;DR

Tubo tackles the challenge of reliable network DM forecasting in the presence of bursts and variable traffic patterns by introducing a tailored ML framework with a Burst Processor and uncertainty-aware Model Selector. It combines burst-aware preprocessing, multiple model types, and calibrated uncertainty via MC Dropout to select the most reliable predictor for each input, achieving up to 4x improvements in DM forecast accuracy and up to 94% burst-occurrence forecasting accuracy. The framework also demonstrates significant downstream benefits for proactive traffic engineering, delivering up to 9x throughput gains over reactive routing and 3x gains over the best existing proactive method. These results, validated on three real-world DM datasets, highlight Tubo’s potential to enhance network performance and reliability in dynamic environments, with an emphasis on practical deployment considerations and scalability.

Abstract

Traffic forecasting based network operation optimization and management offers enormous promise but also presents significant challenges from traffic forecasting perspective. While deep learning models have proven to be relatively more effective than traditional statistical methods for time series forecasting, their reliability is not satisfactory due to their inability to effectively handle unique characteristics of network traffic. In particular, the burst and complex traffic patterns makes the existing models less reliable, as each type of deep learning model has limited capability in capturing traffic patterns. To address this issue, we introduce TUBO, a novel machine learning framework custom designed for reliable network traffic forecasting. TUBO features two key components: burst processing for handling significant traffic fluctuations and model selection for adapting to varying traffic patterns using a pool of models. A standout feature of TUBO is its ability to provide deterministic predictions along with quantified uncertainty, which serves as a cue for identifying the most reliable forecasts. Evaluations on three real-world network demand matrix (DM) datasets (Abilene, GEANT, and CERNET) show that TUBO significantly outperforms existing methods on forecasting accuracy (by 4 times), and also achieves up to 94% accuracy in burst occurrence forecasting. Furthermore, we also consider traffic demand forecasting based proactive traffic engineering (TE) as a downstream use case. Our results show that compared to reactive approaches and proactive TE using the best existing DM forecasting methods, proactive TE powered by TUBO improves aggregated throughput by 9 times and 3 times, respectively.
Paper Structure (17 sections, 1 equation, 7 figures, 4 tables, 2 algorithms)

This paper contains 17 sections, 1 equation, 7 figures, 4 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Problem Statement
  4. Demand Matrix (DM) Forecasting
  5. Characteristics of Demand Matrix Data
  6. Challenges for Demand Matrix Forecasting
  7. Tubo DM Forecasting Framework
  8. Overview
  9. Data Preprocessing
  10. Burst Occurence Forecasting
  11. Model Selection
  12. Evaluation
  13. Methodology
  14. Accuracy of Demand Matrix Forecasting
  15. TUBO Ablation Study
  16. ...and 2 more sections

Figures (7)

  • Figure 1: Tubo framework illustrated.
  • Figure 2: MAE comparison among DM forecasting methods.
  • Figure 3: MAE with synthetic TMGen generated DM data.
  • Figure 4: Impact of burst clipping on MAE. Light-coloured bars show forecasting using raw data, and dark-coloured bars show forecasting with burst clipped.
  • Figure 5: Impact of normalization techniques on MAE with different forecasting methods and DM datasets.
  • ...and 2 more figures