Correlation-Aware User Selection for Multi-User Precoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pre-coding methods for multi-user, multi-antenna transmitters face performance degradation with increasing user correlation and high computational complexity, particularly in massive MIMO systems.
Innovation Solution
A method for reducing the number of users by determining spatial correlation values and discarding users with high correlation, using threshold values specific to different pre-coding techniques, followed by applying linear and non-linear pre-coding based on user groups.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If non-linear pre-coding is applied to address user correlation, then pre-coding performance is improved, but computational complexity increases
Solution Approach 1:
The patent segments users into different groups based on their correlation characteristics. Users are divided into a first group with lower correlation (using linear pre-coding) and a second group with higher correlation (using non-linear pre-coding). This segmentation allows the system to apply complex non-linear pre-coding only where necessary, rather than to all users, thereby improving performance for correlated users while limiting the overall computational complexity increase.
Solution Approach 2:
The patent applies different pre-coding strategies to different user groups based on their local correlation characteristics. The first group of users experiences linear pre-coding with lower complexity, while the second group receives non-linear pre-coding with higher performance. This local differentiation optimizes the balance between performance and complexity for each user group according to their specific channel conditions.
2Device complexity
If linear pre-coding is used for multi-user transmitters, then computational complexity is reduced, but performance deteriorates with increasing user correlation
Solution Approach 1:
The patent segments users into two groups based on correlation thresholds. The first group, with users exhibiting lower correlation, is served using computationally efficient linear pre-coding. The second group, with users showing higher correlation, is served using non-linear pre-coding to maintain performance. This segmentation ensures that linear pre-coding is used where it is effective, maintaining low complexity, while non-linear pre-coding is applied only where performance would otherwise deteriorate.
Solution Approach 2:
The patent changes the pre-coding parameter (linear vs. non-linear) based on the correlation parameter of user groups. By monitoring correlation values and adjusting the pre-coding strategy accordingly, the system adapts to maintain optimal performance while managing computational complexity. This parameter-based adaptation allows dynamic optimization of the complexity-performance tradeoff.
3Device complexity
If user dropping is applied to reduce correlation, then pre-coding complexity is reduced, but system capacity is compromised
Solution Approach 1:
Instead of completely dropping users to reduce correlation, the patent applies partial action by selectively applying non-linear pre-coding to specific user groups that exhibit high correlation. This partial application of non-linear pre-coding addresses the correlation issue only where necessary, rather than requiring user dropping, thereby maintaining system capacity while still reducing overall pre-coding complexity compared to universal non-linear pre-coding.
Data Source
Figure 1
Figure 2~6
Figure 7
AI summary
A pre-coding method is disclosed of a multi-user, multi-antenna transmitter configured for transmission of respective signals to a plurality of users. The method comprises determining (for each pair of users of the plurality) a spatial correlation value for the pair of users, and determining a first correlation threshold value based on the number of users of the plurality. The method also comprises determining whether or not one or more pairs of users has a spatial correlation value that falls on or above the first correlation threshold value. When one or more pairs of users has a spatial correlation value that falls on or above the first correlation threshold value, the method comprises discarding one user comprised in at least one of the one or more pairs to provide a reduced plurality of users, and updating the first correlation threshold value based on the number of users of the reduced plurality. The discarding and updating steps are repeated until none of the spatial correlation values falls on or above the updated first correlation threshold value. Then, the method comprises generating the respective signals for transmission to the users of the reduced plurality by application of pre-coding. Corresponding apparatus, transmitter, base station and computer program product are also disclosed.