Hybrid Machine Learning Models for Fault Detection in 5G Base Station Hardware

The rapid deployment of fifth-generation (5G) wireless communication networks has introduced complex base station architectures consisting of advanced radio units, massive MIMO antennas, power amplifiers, remote radio heads, and edge computing modules. These components operate under demanding conditions, making fault detection a critical requirement for maintaining network reliability and quality of service (QoS). Traditional fault detection methods rely on threshold-based monitoring and manual inspection, which are often inadequate for identifying complex hardware failures in real time. This paper presents a comprehensive study of hybrid machine learning (ML) models for fault detection in 5G base station hardware. The proposed approach combines supervised and unsupervised learning techniques to enhance detection accuracy while minimizing false alarms. Experimental analysis demonstrates that hybrid models outperform conventional machine learning algorithms in detecting hardware anomalies such as amplifier degradation, antenna faults, overheating, and power supply failures. The study highlights the potential of artificial intelligence-driven predictive maintenance in next-generation telecommunications infrastructure.

Cited via DOI — this article is permanently archived on Zenodo.