Neural Network MU-MIMO User Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for user selection in multi-user multiple-input multiple-output (MU-MIMO) communication face high computational complexity and suboptimal performance due to exhaustive search and predefined threshold-based approaches, especially in scenarios with highly correlated users, leading to reduced sum-rate and increased outage probability.
Innovation Solution
A neural network is trained using channel correlation metrics and performance metrics to determine optimal user selection, reducing computational complexity and eliminating the need for predefined thresholds, thereby improving user selection for MU-MIMO communication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If exhaustive search is applied to find optimal user selection, then performance metric (sum-rate) is improved, but computational complexity increases extremely high
Solution Approach 1:
The patent transforms the user selection problem from an exhaustive search over all possible user combinations into a classification problem where channel correlation metrics serve as input parameters. The neural network learns to map these parameters directly to optimal user selection decisions, changing the problem parameters from combinatorial space to metric-based space, thereby reducing computational complexity while maintaining performance.
Solution Approach 2:
The patent replaces the mechanical exhaustive search process with a neural network-based classification system. Instead of systematically evaluating all possible user combinations through computational iteration, the trained neural network directly predicts optimal user selections based on channel correlation metrics, substituting the mechanical search process with an intelligent classification mechanism.
2Device complexity
If spatial correlation evaluation with predefined threshold is used, then computational complexity is reduced, but performance metric (sum-rate) deteriorates and inferiority compared to optimal strategy increases
Solution Approach 1:
The patent changes the decision parameter from a fixed predefined correlation threshold to dynamic channel correlation metrics processed by a neural network. Instead of comparing correlation values against a static threshold, the system uses multiple channel correlation metrics as inputs to a trained classifier that determines user selection based on learned patterns, adapting decisions to specific channel conditions rather than applying uniform threshold rules.
Solution Approach 2:
The patent performs preliminary training of the neural network using exhaustive search results to establish optimal decision boundaries. By pre-learning the relationship between channel correlation metrics and optimal user selections through training data generated from exhaustive search, the system captures optimal selection patterns in advance, enabling fast inference without real-time exhaustive search while maintaining performance.
3Ease of operation
If spatial correlation evaluation with predefined threshold is applied, then process simplicity is improved, but processing resources are consumed for finding suitable threshold values through simulations and measurements
Solution Approach 1:
The patent performs the computationally intensive threshold optimization work in advance during the neural network training phase. By using exhaustive search to generate training data and train the network beforehand, the system transfers the computational burden from runtime operations to offline training, making the actual user selection process simple and resource-efficient while maintaining optimality.
Solution Approach 2:
The patent creates a trained neural network model that copies the optimal decision-making behavior learned from exhaustive search results. Instead of repeatedly performing exhaustive search or threshold optimization, the system uses the trained network as a copy of the optimal strategy that can be applied rapidly to new channel conditions without consuming additional processing resources for threshold finding.
Data Source
AI summary
A method is disclosed of training a neural network to select users for multi user multiple-input multiple-output (MU-MIMO) communication from a set of potential users. The method comprises providing (to the neural network) a plurality of training data sets, each training data set comprising input data corresponding to a channel realization and output data corresponding to an optimal user selection for the channel realization, and controlling the neural network to analyze the plurality of training data sets to determine a branch weight for each association between neurons of neighboring layers of the neural network, wherein the branch weight is for provision of the output data responsive to the input data. A related method of selecting users for MU-MIMO communication from a set of potential users comprises providing (to a neural network trained as specified above) input data corresponding to an applicable channel, receiving (from the neural network) output data comprising a user selection indication, and selecting users based on the user selection indication. Corresponding apparatuses, neural network, network node and computer program product are also disclosed.