AI Channel Link Adaptation via Dimensionality Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing link adaptation techniques in 5G wireless communication networks face challenges in accurately characterizing channel conditions across frequency subcarriers, leading to suboptimal modulation and coding scheme (MCS) selections, which result in throughput losses and reliability issues.
Innovation Solution
The implementation of a low-complexity AI-based system that utilizes dimensionality reduction techniques to generate model input values from signal quality values corresponding to frequency subcarriers, allowing for real-time AI-based classification of optimal MCS parameters, thereby enhancing link adaptation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional link adaptation techniques are used to select MCS based on channel conditions, then the system can maintain basic communication functionality, but the channel conditions cannot be accurately characterized across frequency subcarriers leading to suboptimal MCS selections and throughput losses
Solution Approach 1:
The patent segments the channel condition characterization by frequency subcarriers and uses separate AI models for different subcarrier groups. The dimensionality reduction component processes signal quality values from multiple subcarriers independently, and the classification component selects MCS based on localized channel conditions rather than using a single aggregate metric, thereby improving both measurement precision and productivity.
Solution Approach 2:
The patent introduces an intermediary dimensionality reduction component that transforms the raw signal quality values from frequency subcarriers into reduced-dimensional model input values. This intermediary layer captures the essential channel condition characteristics while reducing computational complexity, enabling accurate channel characterization without sacrificing throughput performance.
2Productivity
If AI-based classification is applied to select optimal MCS in real-time, then link adaptation performance is enhanced and throughput is improved, but the system complexity increases
Solution Approach 1:
The patent extracts the essential channel condition information from the complex raw signal quality values through dimensionality reduction. By taking out only the critical features needed for MCS selection and discarding redundant information, the system achieves real-time AI-based classification with reduced computational complexity, thereby improving throughput without proportionally increasing system complexity.
Solution Approach 2:
The patent transforms the high-dimensional signal quality value space into a lower-dimensional model input space through dimensionality reduction techniques. This dimensional transformation preserves the essential channel condition information while significantly reducing the computational burden on the AI classification model, enabling real-time operation with manageable system complexity.
3Device complexity
If dimensionality reduction is applied to process signal quality values from frequency subcarriers, then the complexity of AI-based classification is reduced, but the number of model input values decreases
Solution Approach 1:
The patent changes the parameters of the input data through dimensionality reduction, transforming the original signal quality values into reduced-dimensional model input values. This parameter transformation consolidates multiple subcarrier measurements into a smaller set of representative features, reducing system complexity while maintaining sufficient information for accurate MCS selection.
Data Source
AI summary
A method for low-complexity artificial intelligence-based channel link adaptation includes converting, by a device including a processor, a first vector of signal quality values corresponding to respective frequency subcarriers utilized by a cell of a communication network into a second vector of model input parameters. The first vector has a first size that is smaller than a second size of the second vector. The method further includes selecting, by the device, a modulation and coding scheme based on applying the second vector to a machine learning model. The method additionally includes facilitating, by the device, conducting a transmission from the cell using the modulation and coding scheme, resulting in the transmission having a first throughput that is higher than a second throughput associated with the cell before the facilitating.