Machine Learning – Based Predictive Modeling for 5G RAN: A Study on Traffic Prediction and Resource Allocation

Abstract

A developing area of computer algorithms called machine learning aims to mimic human intelligence by picking up knowledge from its surroundings. In the modern era of so-called big data, they are regarded as the workhorse. Machine learning techniques have been effectively used in a variety of sectors, including biomedical and medical applications, computer vision, aerospace engineering, pattern identification, finance, entertainment, and computational biology. The operators have been working together globally to improve the radio access network (RAN) architecture, building on the ideas of intelligence and openness. Building an operator-defined RAN architecture (as well as related interfaces) on open hardware that offers intelligent radio control for wireless networks older than fifth generation (5G) is the goal. Innovative radio technologies are envisioned for ultra-dense deployment with enhanced coverage and faster data rates in next-generation 5G wireless networks. However, there are serious issues with network energy usage when super dense 5G networks with comparatively smaller cells are deployed. In order to lower greenhouse gas emissions and operators’ energy bills, new green cloud radio access networks (C-RANs) provide energy-efficient cellular operations. The dynamics of cellular traffic are important for effective network energy management.