Behind the Scenes: An Explainable Artificial Intelligence (XAI) on the Service Classification of the 5G/B5G Network

Abstract

Fifth generation and beyond (5G/B5G) is part of the next generation of mobile technology, which is expected to offer more capacity and faster speeds than the previous generation Long-Term Evolution (LTE) network. These features enable 5G/B5G to offer users a wide range of services, including smart cities, entertainment and multimedia, healthcare and mission-critical applications that benefit the entire economy and communities. However, due to growing demand, telecom service providers need to provide better Quality of service (QoS) on their networks to meet user expectations. Therefore, it is necessary to implement service classification that enables service providers to select the right network slices for each service to improve network QoS. Previous studies have predicted the classification of 5G/B5G services using traditional machine learning algorithms such as random forest and decision tree, achieving good accuracy. However, prediction based on black-box machine learning models was not transparent enough, so mobile service providers could not understand the reliability and interpretation of the results. Therefore, this article aims to model the system performance of 5G Key Performance Indicators (KPI) and extend the experiment to Explainable Artificial Intelligence (XAI) by using the Explainable AI to look behind the scenes of the chosen black-box models for service classification of 5G/B5G networks, as this is crucial for telecom service providers to make critical investment decisions around customer and employee experience. The results from the experiments show that the random forest has the highest confidence towards class Label 3 with the probability of 35 percent.

Ähnliche Arbeiten