5G Technology, 2nd Edition

21Machine Learning for 5G System Optimization

Riku Luostari¹, Petteri Kela¹, Mikko Honkala², Dani Korpi², Janne Huttunen², and Harri Holma¹

¹Nokia, Finland

²Nokia Bell Labs, Finland

CHAPTER MENU

• Introduction
• Motivation
• Model Training and Inference in Wireless Systems
• Machine Learning Categories
• Key Algorithm Techniques
• Machine Learning for 5G Wireless Systems
• Channel State Information (CSI) Improvement and Channel Prediction
• Deep Neural Network-Based Receivers and DeepRx
• Pilotless OFDM
• Massive MIMO, Beamforming, and DeepTx
• Beam Tracking for mmWaves
• Channel Coding
• MAC Scheduler and Radio Resource Management
• Learned Communication Protocols
• Network Planning and Optimization
• Network Operations
• Network Security
• Positioning
• Challenges
• Scalability
• Uncertainty
• Time Criticality and Computational Requirements
• Standardization and Specifications Impact
• Summary
• References

21.1 Introduction

This chapter explores the application of machine learning (ML) in wireless systems, primarily focusing on opportunities in 5G-Advanced, although not exclusively so. The early sections deliver an introduction, highlighting the role of ML in the wireless technology domain, and an overview of training and inference of (ML) models in wireless systems is discussed. Various categories of ML and selected key algorithmic techniques are introduced.

The subsequent sections are dedicated to ML for wireless systems. A broad spectrum of use case examples are explored, such as signal linearization, deep neural ...