CSI Feedback Model Switching for Uplink-Limited Massive MIMO
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CSI feedback methods in MIMO technology, particularly in scenarios with massive MIMO and AI/ML applications, result in inefficient use of uplink resources due to the need to omit all CSI content when resources are insufficient, leading to waste and suboptimal performance.
Innovation Solution
Implement a second model to generate and transmit second CSI information when parts of first CSI need to be omitted, allowing for more efficient use of uplink resources and improving 5G and 6G wireless communication performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional codebook-based CSI feedback methods are used in massive MIMO systems, then channel state information can be fed back to the network device, but the feedback amount becomes very large due to the huge number of antennas, leading to inefficient use of uplink resources
Solution Approach 1:
The patent segments the CSI feedback process into two parts: first CSI generated by a first model and second CSI generated by a second model. This segmentation allows the system to transmit only the most critical CSI information (second CSI) when uplink resources are limited, rather than transmitting all CSI data, thus reducing feedback resource consumption while maintaining essential channel state information
Solution Approach 2:
The patent extracts and prioritizes the most important CSI information by using a second model to generate condensed second CSI from the original first CSI. This extraction process identifies and transmits only the critical channel state parameters needed for effective downlink transmission, eliminating redundant information and reducing overall feedback overhead
2Ease of operation
If all CSI content is omitted when uplink resources are insufficient, then resource allocation is simplified, but uplink resources are wasted and performance becomes suboptimal
Solution Approach 1:
The patent implements a dynamic CSI feedback mechanism where the system adaptively switches between transmitting first CSI, second CSI, or both, based on available uplink resources. When resources are sufficient, both CSI types are transmitted for optimal performance; when resources are limited, only the essential second CSI is transmitted. This dynamic adaptation resolves the contradiction by maintaining both operational simplicity and transmission efficiency across different resource conditions
Solution Approach 2:
The patent changes the parameter of CSI feedback quantity by introducing a second model that generates compressed second CSI with reduced bitwidth. This parameter change allows the system to adjust feedback granularity - transmitting full first CSI when resources allow, or switching to condensed second CSI when resources are constrained, thereby maintaining productivity while preserving ease of resource allocation
3Productivity
If a second model is introduced to generate condensed CSI, then uplink resource utilization improves, but device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training the second model offline to learn the essential characteristics of CSI that are most critical for downlink transmission. This pre-processing step compresses the CSI representation in advance, so that during actual transmission, the system only needs to invoke the pre-trained second model for rapid generation of condensed second CSI. This preliminary preparation reduces real-time computational complexity while maintaining high resource utilization efficiency
Data Source
AI summary
Channel state information transmission and reception methods and apparatuses and a communication system. The transmission apparatus is applicable to a terminal equipment and includes: processor circuitry configured to generate first information of second channel state information (CSI) by using a second model when the terminal equipment determines that at least a part of first CSI needs to be omitted, the at least a part of the first CSI being generated based on a first model; and a first transmitter configured to transmit the second CSI and/or information on the second model to a network device.