Sub-band correlation computation for massive MIMO user pairing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Massive MIMO RANs, existing methods for user pairing and correlation matrix computation are inefficient due to the use of a single wideband correlation metric, leading to sub-optimal MU MIMO performance and increased computational complexity as the number of RRC connected users grows, which can result in sub-optimal MU MIMO performance and high CPU resource requirements.
Innovation Solution
Implementing sub-band correlation computation among scheduled users in a gNB to efficiently pair users across groups of PRBs, where the correlation matrix is computed and shared for each sub-band, and accommodating more RRC connected users by scheduling SRS over multiple groups with shared correlation matrices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single wideband correlation metric is used for user pairing, then the implementation is simple, but the MU MIMO performance becomes sub-optimal
Solution Approach 1:
The patent divides the wideband correlation computation into multiple sub-band correlation computations. Instead of computing a single correlation matrix across the entire bandwidth, the system computes separate correlation matrices for different sub-bands (e.g., low, mid, high frequency sub-bands). This segmentation allows the system to capture frequency-selective channel characteristics more accurately, improving user pairing decisions and MU MIMO performance while keeping individual sub-band computations manageable.
2Productivity
If the number of RRC connected users increases, then more users can be served, but the computational complexity and CPU resource requirements increase
Solution Approach 1:
The patent segments the set of RRC connected users into multiple SRS (Sounding Reference Signal) groups. Each group is processed independently for correlation matrix computation. This grouping strategy reduces the computational complexity from O(N²) for all users to O(k*N²/m) where k is the number of groups and m is the average group size, enabling the system to handle larger numbers of connected users with available CPU resources.
Solution Approach 2:
The patent applies different correlation matrix computation strategies to different user groups based on their channel characteristics and scheduling requirements. By computing correlation matrices locally for each SRS group rather than globally for all users, the system optimizes computational resources while maintaining accurate correlation information for each group's specific context.
3Measurement precision
If sub-band correlation computation is implemented, then the correlation profile accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent divides the frequency band into multiple sub-bands and computes correlation matrices for each sub-band separately. This segmentation provides more accurate correlation profiles by capturing frequency-selective channel characteristics that would be averaged out in a wideband computation. The increased accuracy in user pairing decisions compensates for the additional computational effort required.
Solution Approach 2:
The patent changes the computational parameters by computing correlation matrices at different frequency resolutions (sub-band level) rather than a single wideband level. This parameter change enables the system to adapt the correlation computation granularity to match the frequency selectivity of the channel, improving measurement precision while managing computational complexity through selective sub-band processing.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
System and methods for implementations of sub-band correlation computation among all scheduled users to efficiently pair users across any group of Physical Resource Blocks.