Uplink IRC Split in Massive MIMO Fronthaul
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In massive MIMO systems, the increasing number of antennas leads to a proportional increase in fronthaul capacity, resulting in high costs. Existing solutions, such as enhanced CPRI, only partially address this issue by moving frequency-domain beamforming from the BBU to the RU, reducing fronthaul capacity but not eliminating the need for high-capacity transport.
Innovation Solution
The proposed IRC split method distributes the computational complexity of interference rejection combining (IRC) between the radio unit (RU) and the distributed unit (DU), reducing the number of uplink data streams to the number of user layers. This is achieved by calculating the first part of beamforming weights in the RU and the second part in the DU, effectively implementing an IRC receiver.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the number of antennas is increased in massive MIMO systems, then the spectrum efficiency and cell capacity are improved, but the fronthaul capacity requirement increases proportionally, resulting in high costs
Solution Approach 1:
The baseband processing function is segmented into two parts: frequency-domain beamforming (first part of IRC) is performed at the RU, while the second part of IRC is performed at the DU. This segmentation allows the RU to process signals locally and transmit only processed data streams to the DU, reducing fronthaul capacity requirements while maintaining massive MIMO performance
Solution Approach 2:
The patent introduces an intermediate processing stage at the RU that performs partial IRC and frequency-domain beamforming. This intermediary processing reduces the dimensionality of data transmitted over the fronthaul interface, acting as a mediator between the antenna array and the central baseband processing unit
2Ease of operation
If all IRC processing is performed at the RU, then the processing is simplified at the RU side, but the fronthaul capacity requirement remains high because N-dimensional data streams must be transmitted
Solution Approach 1:
The IRC processing is divided into two parts: the first part (HHQ^-1) is performed at the RU to reduce data dimensionality, and the second part is performed at the DU. This segmentation enables the RU to perform simplified processing while transmitting only K-dimensional data streams (where K is the number of user layers, much smaller than N antennas) to the DU
3Reliability
If the number of antennas is increased, then the interference rejection capability is improved, but the computational complexity for calculating IRC coefficients increases due to N×N matrix inversion
Solution Approach 1:
The computationally intensive IRC calculation is segmented and distributed: the RU performs the first part (HHQ^-1) which involves N×K operations, and the DU performs the second part. This distribution reduces the peak computational complexity at any single location compared to performing full N×N matrix inversion centrally
Solution Approach 2:
The RU performs preliminary IRC processing (first part) before transmitting data to the DU. This preliminary action reduces the dimensionality of subsequent processing at the DU and prepares data in advance, reducing overall computational complexity
Data Source
AI summary
A method by a RU node for performing frequency-domain beamforming for communication between a base station (BS) and UEs in a network using a multiple antenna system, the BS including a DU node connected to the RU node, the method including: obtaining uplink signals including K user-layer signals overlaid with interference signals and noise as received at N antennas from a number of UEs; determining: a channel estimation matrix H of wireless communication channels between a number of UEs and N antennas; an estimate of an Interference plus Noise covariance matrix Q based on H and other channel information; a first part beamforming weights, BFWs; an effective channel matrix Heff based on H and the first part BFWs; and intermediate uplink signals having K components and based on the uplink signals and the first part BFWs; and sending Heff and the intermediate uplink signals towards the DU node.