Tidal Traffic Prediction for Reliable Optical Network Orchestration in Industry 5.0
Igor Kardush, Sejeong Kim, Elaine Wong · 2023
Industry 5.0 is expected to involve humans and machines collaborating on the factory floor, relying on an ultra-reliable and low latency human-to-machine (H2M) communication system to support interaction between humans and cobots. Optical enterprise networks are a natural choice for deploying Industry 5.0, we explore six candidate optical enterprise network architectures in their most cost-effective topologies for Industry 5.0. We propose a tidal traffic prediction and network orchestration scheme that utilizes a recurrent neural network (RNN) based on Long Short-Term Memory (LSTM) to forecast enterprise tidal traffic and proactively enable protection paths to improve network reliability. A sensitivity analysis is used to optimize the parameters of the LSTM traffic predictor, resulting in a rapid machine-learning convergence. The scheme is evaluated to the six candidate optical enterprise network architectures resulting in improvement of network connection availability when tidal traffic prediction is used to orchestrate protection paths on the enterprise optical network.