Machine Learning 
in Wireless Networks!
Wireless machine learning (ML) will function as an overlay in the early phases of 5G.
This overlay machine learning aims to enhance current features like beam management📡, energy conservation🔋, and quality of service (QoS) enhancements📈. It can be applied separately on the device📱 or in the network🌐. The device or infrastructure vendor is responsible for implementing wireless ML as proprietary functions, including model development and management💼.
Only one-sided models—in which inference is done on the device📱 or the network🌐—are supported by overlay machine learning.
Like in 5G NR Release 16 and beyond, these capabilities also rely on proprietary and standardized data collecting:bar_chart:, which is utilized as training input for self-organizing networks and minimization of drive tests (SON/MDT)
.
3GPP: Rel17-RAN3 & Rel18-RAN1 Focus on AI/ML Studies in RAN networks