MU-MIMO User Selection via Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current user selection algorithms in 5G MU-MIMO systems, such as CUS and DDPG, are computationally intensive and inefficient, leading to high CPU occupancy and resource waste due to their complexity and failure to consider multi-user channel quality indicators and time dimension in scheduling, which affects network spectrum efficiency and throughput.
Innovation Solution
A modified DDPG algorithm integrated with AI/ML techniques that uses a reinforcement learning framework to reduce computational complexity by training a user selection model with historical data, considering multi-user channel quality indicators and scheduling priorities, and deploying it in a real-time wireless protocol stack to optimize user selection and resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional user selection algorithms (CUS, DDPG) are used in 5G MU-MIMO systems, then user selection can be performed, but computational complexity is high and CPU occupancy is high
Solution Approach 1:
The patent pre-calculates and stores channel quality indicators (CQIs) for different resource blocks before the actual scheduling decision. By performing this computation in advance rather than in real-time during scheduling, the system reduces the computational burden during the critical scheduling moment while maintaining accurate user selection based on multi-user channel conditions.
Solution Approach 2:
The patent divides the user selection process into multiple stages: first identifying candidate users based on basic criteria, then filtering and ranking them using pre-calculated CQIs. This segmentation of the selection process allows complex computations to be distributed and optimized, reducing overall computational complexity while improving selection efficiency.
2Productivity
If traditional scheduling algorithms are used, then resource allocation can be performed, but network spectrum efficiency is reduced due to not considering multi-user channel quality indicators
Solution Approach 1:
The patent implements a feedback mechanism where channel quality indicators are continuously measured, stored, and used to inform subsequent scheduling decisions. The system feeds back the pre-calculated CQIs to the scheduler, enabling it to make informed decisions about which users to select and which resource blocks to allocate, thereby optimizing network spectrum efficiency based on actual channel conditions.
Solution Approach 2:
The patent changes the scheduling parameter from simple correlation-based metrics to multi-user channel quality indicators (CQIs) that capture the actual channel conditions between base station and users. By using these enhanced parameters that reflect true channel quality, the system achieves better spectrum efficiency and user selection accuracy.
3Productivity
If complex algorithms are used to achieve higher throughput, then throughput may improve, but CPU occupancy increases and resources are wasted
Solution Approach 1:
The patent uses lightweight, pre-calculated channel quality indicators instead of performing heavy real-time matrix computations for each scheduling decision. These pre-computed CQIs act as simplified proxies that enable throughput optimization without requiring continuous expensive CPU-intensive calculations, thus reducing CPU occupancy while maintaining high throughput performance.
Solution Approach 2:
The system performs computationally intensive channel quality analysis in advance and stores the results. During actual scheduling, it only needs to retrieve and compare these pre-computed values, avoiding repeated heavy computations. This preliminary action significantly reduces real-time CPU occupancy while enabling sophisticated user selection that maintains high throughput.
Data Source
AI summary
Processing circuitry for a communication station configured to facilitate multi-user multiple-input multiple output (MU-MIMO) service. The processing circuitry can perform a multi-user selection for data transmission on a shared radio resource from a plurality of User Equipments (UEs). The processing circuitry selects one or more of the plurality of candidate UEs in time domain based on time-domain scheduling algorithm, obtain historical throughput data and input for each selected UE. The input includes a channel state indicator including a single-user channel quality indicator (SU-CQI), a precoding matrix indicator (PMI), rank indicator, and a channel state matrix. A trained reinforcement learning agent (RL agent) using the obtained input infers a rating score for each of the plurality of UEs. The processing circuitry schedules the one or more the UEs for transmission respectively on the plurality of radio resources based on the plurality of score ratings and allocate the plurality of radio resources.