Machine Learning for Large-Scale Optimization in 6G Wireless Networks

Abstract

The sixth generation (6G) wireless systems are envisioned to enable the paradigm shift from "connected things" to "connected intelligence", featured by ultra high density, largescale, dynamic heterogeneity, diversified functional requirements and machine learning capabilities, which leads to a growing need for highly efficient intelligent algorithms. The classic optimization-based algorithms usually require highly precise mathematical model of data links and suffer from poor performance with high computational cost in realistic 6G applications. Based on domain knowledge (e.g., optimization models and theoretical tools), machine learning (ML) stands out as a promising and viable methodology for many complex large-scale optimization problems in 6G, due to its superior performance, generalizability, computational efficiency and robustness. In this paper, we systematically review the most representative "learning to optimize" techniques in diverse domains of 6G wireless networks by identifying the inherent feature of the underlying optimization problem and investigating the specifically designed ML frameworks from the perspective of optimization. In particular, we will cover algorithm unrolling, learning to branch-andbound, graph neural network for structured optimization, deep reinforcement learning for stochastic optimization, as well as endto-end learning for semantic optimization, for solving challenging large-scale optimization problems arising from various important wireless applications. To enable ML implementation in distributed wireless networks across massive number of end devices, federated learning for distributed optimization will further be presented. Through the in-depth discussion, we shed light on the excellent performance of ML-based optimization algorithms with respect to the classical methods, and provide insightful guidance to develop advanced ML techniques in 6G networks. Neural network design, theoretical tools of different ML methods,

New developed Methods used for Data Science Optimization

“…In their study. Yandong Sh et al [20] examined different techniques for "learning to optimize" in 6G wireless networks. They used ML frameworks to identify the main characteristics of the optimization problem in various domains.…”

Optimizing Multicarrier Multiantenna Systems for LoS Channel Charting

“…The next generation of cellular network (6G) standards will most likely be marked by a profound integration and utilization of artificial intelligence and in particular machine learning (ML) methods [1], [2], [3], [4], [5]. Indeed, their ability to learn complex patterns from data makes them suitable for many applications where traditional model-based approaches rapidly reach their limit, especially with the advent of massive multiple input multiple output (MIMO) systems where very large bandwidths and a large number of antennas are considered.…”

A Metaheuristics-Based Hyperparameter Optimization Approach to Beamforming Design

“…The needs for the 6G systems have necessitated the granular optimization of radio resources and the efficient acquisition of network-related data [4]. Due to the huge size, high density, the varied quality of services, and integrated multi-functional cross-layer architecture, 6G optimization problems might be exceedingly complex and time-sensitive, posing many challenges to the development of effective optimization algorithms.…”

“…Due to the huge size, high density, the varied quality of services, and integrated multi-functional cross-layer architecture, 6G optimization problems might be exceedingly complex and time-sensitive, posing many challenges to the development of effective optimization algorithms. Deep learning (DL) has recently been utilized as a disruptive method to solve difficult optimization problems in 6G and to support a number of artificial intelligence services and the Internet of Everything applications [4]. It has also been proven to be a useful tool for dealing with difficult non-convex problems and high-computability concerns owing to its excellent recognition and representation capabilities [5].…”

“…It has also been proven to be a useful tool for dealing with difficult non-convex problems and high-computability concerns owing to its excellent recognition and representation capabilities [5]. To enable a paradigm shift from traditional optimization theory-based approaches for employing more promising DL architectures, DL-based optimization algorithm design aims to achieve near-optimal performance with excellent computing efficiency for challenging large-scale optimization problems in 6G systems [4]. In particular, superior performance, scalability and generalizability, computational efficiency, and robustness are some benefits of using DL for large-scale optimization.…”