New systems such as the 5G/6G mobile network, vehicular networks, the Internet of things, and edge computing, among others, are large-scale, distributed, and support user profiles with dynamic and heterogeneous requirements. These new systems require a dynamic network operation that adapts to the systems’ requirements. Network slicing (NS) is a crucial enabler and feature for these new systems. NS, in summary, virtualizes network resources in a dynamic and customized way concerning the demands of the systems. Network slicing provides systems, operators, and service providers with flexibility in orchestrating and allocating resources with cost, efficiency, and scale gains. However, despite NS’s numerous benefits, the dynamic and efficient orchestration of resources is a critical gap and challenge for designing architectures and systems based on network slicing. The NSML (Energy-Efficient and Edge-Aware Network Slicing with Machine Learning - NSML) project proposes developing resource orchestration models for network slicing. The NSML orchestration models focus on energy efficiency and edge computing in an Internet of Things (IoT) scenario with communication (Publish/Subscribe) (Pub/Sub). The base network slicing architecture of the NSML project is the SFI2 (Slicing Future Internet Architectures) reference architecture. The NSML project adopts the machine learning strategy native to the architecture (ML-native) and sustainably orchestrates resources by optimizing energy efficiency. Slicing for edge computing with IoT uses the Pub/Sub paradigm with machine learning for optimization at scale. In summary, NSML develops state-of-the-art orchestration models for network slicing using energy-efficient native ML with Pub/Sub-style edge computing. 

PPGCOMP UNIFACS supports the NSML project

PPDRU UNIFACS supports the NSML project