Towards ML/AI-based Prediction of Mobile Service Usage in Next-Generation Networks

TL;DR

This paper presents a machine learning approach to predict mobile service usage in next-generation networks, utilizing data from the Network Slice Planner v2 to improve adaptive decision-making and infrastructure performance.

Contribution

It introduces a novel method combining supervised and unsupervised learning for analyzing mobile user behavior using NSP v2 data.

Findings

High accuracy in predicting mobile service consumption

Effective analysis of user behavior patterns

Improved performance of MEC infrastructure

Abstract

The adoption of machine learning techniques in next-generation networks has increasingly attracted the attention of the research community. This is to provide adaptive learning and decision-making approaches to meet the requirements of different verticals, and to guarantee the appropriate performance requirements in complex mobility scenarios. In this perspective, the characterization of mobile service usage represents a funda-mental step. In this vein, this paper highlights the new features and capabilities offered by the "Network Slice Planner"(NSP) in its second version [12]. It also proposes a method combining both supervised and unsupervised learning techniques to analyze the behavior of a mass of mobile users in terms of service consumption. We exploit the data provided by the NSP v2 to conduct our analysis. Furthermore, we provide an evaluation of both the accuracy of the predictor and the performance of the underlying MEC infrastructure.