Massive MIMO Precoding Complexity Reduction via Spatial FFT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high computational complexity of pre-coding matrix computation, particularly matrix multiplication, in massive Multiple-Input Multiple-Output (MIMO) systems hinders real-time implementation due to the large number of antennas and user equipment, with existing methods focusing on reducing matrix inversion complexity but not addressing the more significant O(MK2) complexity of multiplications.
Innovation Solution
The implementation of spatial Fast Fourier Transform (FFT) to convert signal vectors into sparse forms, reducing the complexity of matrix multiplication and facilitating efficient UE grouping by mapping signals to limited angles, thereby simplifying the multiplication process and reducing computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Zero-Forcing beamforming is used to achieve high throughput and performance close to channel capacity, then the system throughput is improved, but the computation complexity of pre-coding matrix increases to O(MK²) for matrix multiplication
Solution Approach 1:
The patent segments the M BS antenna array into N sub-arrays, where each sub-array has M/N antennas. This segmentation transforms the original large-scale pre-coding problem into multiple smaller sub-problems, reducing the matrix multiplication complexity from O(MK²) to O(NK² + NK(M/N)) = O(NK² + K M), which is significantly lower when N is chosen appropriately.
Solution Approach 2:
The patent introduces a new dimension by transforming the pre-coding matrix computation into a two-stage process: first computing sub-array pre-coding matrices independently, then combining them through a specific matrix operation. This dimensional transformation allows the system to bypass the direct O(MK²) computation by operating in a decomposed computational space.
2Use of energy by moving object
If a large number of antennas are used to focus energy into individual UEs, then radiated energy efficiency and throughput are improved, but the complexity of pre-coding matrix computation increases significantly
Solution Approach 1:
The patent divides the large antenna array into multiple sub-arrays, allowing energy focusing to be achieved through coordinated transmission from smaller sub-arrays rather than requiring simultaneous computation across all antennas. This maintains the energy focusing capability while reducing computational burden through the segmented architecture.
Solution Approach 2:
The patent computes pre-coding matrices for sub-arrays separately and then combines them, performing partial computations that are then aggregated. This partial action approach computes only the necessary components (sub-array matrices) rather than the full M×K matrix directly, reducing overall computational complexity while achieving the same energy focusing effect.
3Productivity
If many UEs are scheduled into different groups for service, then system capacity is improved, but the complexity of UE grouping and correlation matrix construction increases
Solution Approach 1:
The patent segments the UE grouping process by associating UEs with specific sub-arrays based on their channel characteristics. This segmentation allows the base station to manage groups of UEs more efficiently by distributing them across sub-arrays, reducing the complexity of constructing and managing correlation matrices for all UEs simultaneously.
Solution Approach 2:
The patent performs preliminary channel estimation and correlation analysis at the sub-array level before final UE grouping. This preliminary action allows the system to pre-identify suitable UE groups for each sub-array, simplifying the overall grouping process and reducing computational complexity by avoiding exhaustive search across all possible UE combinations.
Data Source
AI summary
This invention presents methods for using spatial FFT to reduce the number of computations for generating the pre-coding matrix in a MIMO system comprising reducing the dimension of channel vectors by neglecting entries whose values are significantly smaller or near zero, and to select UEs into a group assigned to the same time and frequency resources.