Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FUSION: A Flexible Unified Simulator for Intelligent Optical Networking

Ryan McCann, Arash Rezaee, Vinod M. Vokkarane

TL;DR

FUSION addresses the lack of an open, feature-rich SD-EON simulator by delivering a modular, Python-based platform that integrates traditional RSA methods with ML and reinforcement learning, reinforced by a GUI, 90% unit test coverage, and SB3 compatibility. The paper demonstrates performance gains through light-segment slicing, XT-aware resource allocation, and RL-based path selection, underscoring reduced blocking and enhanced spectral efficiency. Its open-source nature and thorough documentation aim to accelerate research in SD-EONs and enable widespread collaboration. Overall, FUSION provides a robust, extensible foundation for exploring intelligent, software-defined elastic optical networks in dynamic traffic settings.

Abstract

The increasing demand for flexible and efficient optical networks has led to the development of Software-Defined Elastic Optical Networks (SD-EONs). These networks leverage the programmability of Software-Defined Networking (SDN) and the adaptability of Elastic Optical Networks (EONs) to optimize network performance under dynamic traffic conditions. However, existing simulation tools often fall short in terms of transparency, flexibility, and advanced functionality, limiting their utility in cutting-edge research. In this paper, we present a Flexible Unified Simulator for Intelligent Optical Networking (FUSION), a fully open-source simulator designed to address these limitations and provide a comprehensive platform for SD-EON research. FUSION integrates traditional routing and spectrum assignment algorithms with advanced machine learning and reinforcement learning techniques, including support for the Stable Baselines 3 library. The simulator also offers robust unit testing, a fully functional Graphical User Interface (GUI), and extensive documentation to ensure usability and reliability. Performance evaluations demonstrate the effectiveness of FUSION in modeling complex network scenarios, showcasing its potential as a powerful tool for advancing SD-EON research.

FUSION: A Flexible Unified Simulator for Intelligent Optical Networking

TL;DR

FUSION addresses the lack of an open, feature-rich SD-EON simulator by delivering a modular, Python-based platform that integrates traditional RSA methods with ML and reinforcement learning, reinforced by a GUI, 90% unit test coverage, and SB3 compatibility. The paper demonstrates performance gains through light-segment slicing, XT-aware resource allocation, and RL-based path selection, underscoring reduced blocking and enhanced spectral efficiency. Its open-source nature and thorough documentation aim to accelerate research in SD-EONs and enable widespread collaboration. Overall, FUSION provides a robust, extensible foundation for exploring intelligent, software-defined elastic optical networks in dynamic traffic settings.

Abstract

The increasing demand for flexible and efficient optical networks has led to the development of Software-Defined Elastic Optical Networks (SD-EONs). These networks leverage the programmability of Software-Defined Networking (SDN) and the adaptability of Elastic Optical Networks (EONs) to optimize network performance under dynamic traffic conditions. However, existing simulation tools often fall short in terms of transparency, flexibility, and advanced functionality, limiting their utility in cutting-edge research. In this paper, we present a Flexible Unified Simulator for Intelligent Optical Networking (FUSION), a fully open-source simulator designed to address these limitations and provide a comprehensive platform for SD-EON research. FUSION integrates traditional routing and spectrum assignment algorithms with advanced machine learning and reinforcement learning techniques, including support for the Stable Baselines 3 library. The simulator also offers robust unit testing, a fully functional Graphical User Interface (GUI), and extensive documentation to ensure usability and reliability. Performance evaluations demonstrate the effectiveness of FUSION in modeling complex network scenarios, showcasing its potential as a powerful tool for advancing SD-EON research.

Paper Structure

This paper contains 15 sections, 5 equations, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. System Architecture
  4. Methodology
  5. Algorithms and Techniques
  6. Standard Routing and Spectrum Assignment Algorithms
  7. Crosstalk-Aware Algorithms
  8. Reinforcement Learning Algorithms
  9. Supervised Learning Algorithms
  10. Comparison and Validation
  11. Performance Evaluation
  12. Light Segment Slicing
  13. XT-Aware Algorithms
  14. Reinforcement Learning-Based Path Selection
  15. Conclusion

Figures (5)

  • Figure 1: High-level design of the FUSION architecture.
  • Figure 2: FUSION GUI displaying the USNet Topology.
  • Figure 3: BP vs. Load for Light-Segment Slicing.
  • Figure 4: BP vs. Load for XTAR & XTA-SMCSA.
  • Figure 5: BP vs. Episodes for 750 Erlang.