Multi-User Beam Alignment for Asymmetric mmWave MIMO

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Asymmetric millimeter wave large-scale multiple-input multiple-output (MIMO) systems face high training overhead and complexity in beam alignment due to the need for separate uplink and downlink beam training, which restricts their development and application.

Innovation Solution

A multi-user uplink and downlink beam alignment method using a multi-directional beam codebook, optimized codewords, and a convolutional neural network for predicting optimal beams, reducing training overhead by leveraging partial reciprocity between uplink and downlink domains.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If separate uplink and downlink beam training is performed in asymmetric MIMO systems, then reliable beam alignment can be achieved, but training overhead and system complexity increase significantly

Engineering Contradiction:
Improvebeam alignment reliabilityVSAvoidbeam training complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent combines uplink and downlink beam training into a unified process by utilizing channel reciprocity. The base station performs downlink beam training while simultaneously acquiring uplink channel information through reciprocal relationship, eliminating the need for separate uplink beam training procedures and reducing overall training overhead while maintaining alignment reliability.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The downlink beam training process is designed to serve multiple functions: it establishes downlink beam alignment, acquires channel state information, and enables inference of uplink beam directions through reciprocity. This multi-functional approach reduces the need for dedicated uplink training resources and simplifies the overall beam alignment procedure.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If traditional beam training methods are used in asymmetric MIMO systems, then beam alignment can be achieved, but training overhead increases

Engineering Contradiction:
Improvebeam alignmentVSAvoidtraining overhead
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The base station performs downlink beam training in advance to establish beam alignment and acquire channel information before uplink transmission begins. By preparing the beam alignment information preliminarily through downlink training, the system eliminates the need for additional uplink training overhead, as the beam alignment results are already available through channel reciprocity.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If all-digital multi-beam arrays are used in millimeter wave systems, then beamforming performance is improved, but overhead, complexity and power consumption increase

Engineering Contradiction:
Improvebeamforming gainVSAvoidsystem complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent exploits the asymmetric nature of uplink and downlink channel characteristics in millimeter wave systems. By designing the beam training process to leverage asymmetry and reciprocity relationships, the system achieves effective beam alignment with reduced complexity compared to symmetric approaches that would require identical training procedures for both uplink and downlink.

Inventive Principle:
Principle #4Asymmetry

Data Source

PatentUS11742910B2Multi-user uplink and downlink beam alignment method for asymmetric millimeter wave large-scale MIMO
Publication Date: 2023.08.29 ZHEJIANG UNIV
  • US11742910B2 patent drawing
  • US11742910B2 patent drawing
  • US11742910B2 patent drawing

AI summary

A multi-user uplink and downlink beam alignment method for asymmetric millimeter wave large-scale MIMO includes constructing an all-digital multi-directional beam for multiple-direction probing; performing multiple rounds of downlink beam training from the base station to the UE; controlling the UE to perform a beam decision according to the receiving signals, and determining a target downlink sending-receiving beam pair; performing data processing on the receiving signals to generate training data for predicting an uplink sending narrow beam, and performing training on a preset neural network; performing online real-time signal detection based on the trained network parameters and the receiving signals to predict a target uplink sending narrow beam; and controlling the UE to feedback an index of a target downlink sending narrow beam to the base station, and widening the target downlink sending narrow beam into a target uplink receiving beam according to the feedback index.