Two Use Cases of Machine Learning for SDN-Enabled IP/Optical Networks: Traffic Matrix Prediction and Optical Path Performance Prediction [Invited]
Gagan L. Choudhury, David F. Lynch, Gaurav Thakur, Simon Tse
Abstract
Gagan L. Choudhury, David F. Lynch, Gaurav Thakur, Simon Tse
Abstract
We describe two applications of machine learning in the context of internet protocol (IP)/Optical networks. The first one allows agile management of resources in a core IP/Optical network by using machine learning for short-term and long-term prediction of traffic flows. It also allows joint global optimization of IP and optical layers using colorless/directionless (CD) reconfigurable optical add-drop multiplexers (ROADMs). Multilayer coordination allows for significant cost savings, flexible new services to meet dynamic capacity needs, and improved robustness by being able to proactively adapt to new traffic patterns and network conditions. The second application is important as we migrate our networks to Open ROADM networks to allow physical routing without the need for detailed knowledge of optical parameters. We discuss a proof-of-concept study, where detailed performance data for established wavelengths in an existing ROADM network is used for machine learning to predict the optical performance of each wavelength. Both applications can be efficiently implemented by using a software-defined network controller.
OpenAlex reports 58 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
We describe two applications of machine learning in the context of internet protocol (IP)/Optical networks. The first one allows agile management of resources in a core IP/Optical network by using machine learning for short-term and long-term prediction of traffic flows. It also allows joint global optimization of IP and optical layers using colorless/directionless (CD) reconfigurable optical add-drop multiplexers (ROADMs). Multilayer coordination allows for significant cost savings, flexible new services to meet dynamic capacity needs, and improved robustness by being able to proactively adapt to new traffic patterns and network conditions. The second application is important as we migrate our networks to Open ROADM networks to allow physical routing without the need for detailed knowledge of optical parameters. We discuss a proof-of-concept study, where detailed performance data for established wavelengths in an existing ROADM network is used for machine learning to predict the optical performance of each wavelength. Both applications can be efficiently implemented by using a software-defined network controller.
Key concepts: Computer science, Multiplexer, Computer network, Software-defined networking, Distributed computing, Multiplexing, Telecommunications