5G Network Slicing with Multi-Purpose AI models

Fifth-Generation (5G) communication systems in-tend to meet stringent quality of service requirements such as reliable communication, low latency, providing high data rates with security constraints. The need for programmable solutions to enable the provision of services depending on these requirements can be addressed through Network Slicing to be deployed in 5G networks. In this paper, by simulating realistic user and base station data, artificial intelligence and machine learning techniques are explored for associating users with different types of communication slices. Instead of dealing with a snapshot of the environment, an evolving real-time scenario is adopted where hand overs between cells are taken into account. Neural network-based and random forest-based learning models were developed and tested in the placement of the users to Massive Internet of Things (MIoT), Ultra-Reliable Low-Latency Communications (URLLC), Vehicle to Everything (V2X), and enhanced Mobile Broadband (eMBB) slices.