MLMD Dataset

The dataset is built upon a multi-layer multi-domain (MLMD) network topology of a real testbed. The topology includes an IP layer connected to an Optical Transport Network (OTN) layer and a Dense Wavelength Division Multiplexing (DWDM) layer through FlexEthernet (FlexE) links.  The dataset explores various traffic load and  FlexE physical link (PHY) failure scenarios, making it suitable for simulations and Machine Learning (ML) tasks involving network analysis such as routing optimization and traffic restoration. The dataset is partitioned into three separate subsets: a training set containing 12512 samples, a validation set comprising 2144 samples, and an evaluation set with an additional 1000 samples. Each sample is described by a feature matrix and a label vector.