Selection of Waveform Parameters Using Machine Learning for 5G and Beyond

Abstract

Flexibility is one of the essential requirements in future cellular communications technologies. Providing customized communications solutions for each user and service type cannot be possible without the flexibility in 5G and beyond. Different optimizations need to be done for the flexibility related structures of 5G and beyond systems. In this paper, a novel machine learning (ML) based selection mechanism for the configurable waveform parameters is designed from the flexibility perspective. Moreover, a simulation based dataset generation methodology is proposed for ML systems. Results of computer simulations are presented using the generated dataset.