【综述一览】A Survey of Recent Advances in Optimization Methods for Wireless Communications

I. INTRODUCTION

A. Evolution of Wireless Cellular Communication Systems: From 1G to 6G

B. Central Role of Mathematical Optimization in Wireless Communication System Design

C. Goals of the Paper

Provide an overview of recent advances in mathematical optimization theory and algorithms

Guide the choice and development of suitable algorithms for solving structured optimization problems

Promote the cross-fertilization of ideas in mathematical optimization and wireless communications

D. Structure of the Paper

II. OPTIMIZATION PROBLEMS IN WIRELESS COMMUNICATIONS: STRUCTURES AND CHALLENGES

A. Optimization Problems in Wireless Communications

1) Optimization Problems with Continuous Variables

a) Downlink Beamforming

b) Hybrid Beamforming

c) RIS Beamforming

d) Joint BS Clustering and Beamformer Design

e) Joint Downlink Beamforming and Fronthaul Compression

2) Optimization Problems with Integer Variables

a) MIMO Detection

b) Symbol-Level Precoding

3) Optimization Problems with Mixed Variables

a) Joint Admission Control and Multicast Beamforming

b) Joint Uplink Scheduling and Power Control

4) Remarks

B. Challenges from the Optimization Perspective

Large dimensionality and high nonlinearity

Lack of favorable properties

Mixed-integer variables

C. Structural Properties of Optimization Problems

III. STRUCTURED NONCONVEX OPTIMIZATION

A. Fractional Programming

1) Two FP Transforms

a) Quadratic Transform

b) Lagrangian Dual Transform

2) Application Examples

a) Downlink Beamforming for Sum-Rate Maximization

b) Joint Uplink Scheduling and Power Control for SumRate Maximization

3) Remarks

B. Sparse Optimization

1) Compressed Sensing and Recovery Conditions

2) Application Examples

a) Localized Statistical Channel Modeling

b) Device Activity Detection in mMTC

3) Remarks

C. Proximal Gradient Algorithms

1) PG Algorithm, Interpretation, and Convergence Property

2) Application Examples

a) Massive MIMO Detection

b) Joint BS Clustering and Beamformer Design

3) Remarks

D. Penalty Methods

1) Penalty Methods for a Class of Optimization Problems with Structured Constraints

2) Remarks

E. Duality-Based Algorithms

1) Uplink-Downlink Duality

2) Duality-Based Algorithms

3) Remarks

IV. PROBLEM-SPECIFIC GLOBAL OPTIMIZATION

A. Introduction to B&B and B&C Frameworks

1) B&B Algorithmic Framework

2) B&C Algorithmic Framework

B. SDRs for a Class of CQPs

C. Mixed Monotonic Programming

D. Valid Cuts for Mixed-Integer Problems

V. DISTRIBUTED OPTIMIZATION AND FEDERATED LEARNING

A. Decomposition Methods

1) Dual Decomposition, ADMM, and Their Variants

2) Application Example

B. Federated Learning in Wireless Edge Networks

1) FEDAVG Algorithm

2) Performance Analysis

3) Improved FL Algorithms

VI. OPTIMIZATION VIA LEARNING

A. Black-Box Based Approaches

B. Unfolding-Based Approaches

VII. LEARNING-BASED OPTIMIZATION WITHOUT EXPLICIT CHANNEL ESTIMATION

A. Channel Modeling and CSI Estimation

B. Model-Free Optimization

1) Scheduling and Power Control Using Geographic Information

2) Beamforming and RIS Reconfiguration with Implicit Channel Estimation

C. Neural Network Architecture Considerations

VIII. OPEN PROBLEMS AND RESEARCH DIRECTIONS

A. Open Problems

B. Research Directions

1) Distributed Signal Processing and Optimization for Extremely Large-Scale Antenna Array (ELAA) Systems

2) Quantum Optimization and ML

IX. CONCLUSION