ML-Based Channel State Information Feedback Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a lack of established techniques for effectively applying and controlling machine learning technology in wireless communication within mobile communication systems, particularly in the 5G and future generations of mobile communication systems.
Innovation Solution
The implementation of a machine learning technology in a mobile communication system involves configuring communication apparatuses, such as User Equipment (UE) and base stations, with functional blocks for model learning and inference, enabling the derivation and use of learned models for controlling transmission and reception of control data related to channel state information feedback, allowing for efficient CSI feedback and overhead reduction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning technology is applied to wireless communication for channel state information feedback, then measurement precision and resource utilization are improved, but device complexity and overhead increase
Solution Approach 1:
The patent segments the machine learning functionality into separate functional blocks: a model learning unit that derives learned models from training data, and a model inference unit that applies these models to infer channel state information. This segmentation allows the complex ML processing to be organized into manageable, specialized components, reducing overall device complexity while maintaining measurement precision.
Solution Approach 2:
The patent implements preliminary action by performing model learning during periods when inference is not required, deriving learned models in advance and storing them for later use. This allows the complex learning process to occur beforehand, so that during actual communication operations, only the simpler inference process needs to run, reducing real-time computational complexity while maintaining high measurement precision.
2Measurement precision
If machine learning model learning and inference are performed in communication apparatuses, then channel state information feedback accuracy is improved, but power consumption increases
Solution Approach 1:
The patent implements periodic action by alternating between model learning operations and model inference operations. The system performs model learning during specific periods when training data is available and computational resources are abundant, then switches to inference mode during communication operations. This periodic alternation allows the system to maintain high measurement precision while managing power consumption by not continuously running the computationally intensive learning process.
Solution Approach 2:
By performing model learning in advance during preliminary action phases, the system prepares learned models that can be reused during inference phases. This eliminates the need to continuously perform learning operations during communication operations, significantly reducing ongoing power consumption while maintaining the precision benefits of machine learning-based channel state information feedback.
3Productivity
If learned models are used for channel state information inference, then productivity and resource utilization are improved, but loss of information may occur during model learning and control data transmission
Solution Approach 1:
The patent implements feedback mechanisms where the system monitors the performance and accuracy of learned models during inference operations. This feedback allows the system to detect when information loss or degradation occurs, and can trigger re-learning or model updates to restore accuracy. The feedback loop ensures that productivity gains from using learned models do not come at the cost of significant information loss.
Solution Approach 2:
The patent uses copying by transmitting control data that represents learned models between communication apparatuses. Instead of transmitting raw channel state information, the system transmits compressed model representations that can be reconstructed at the receiving end. This copying approach reduces transmission overhead and potential information loss while maintaining productivity benefits.
Data Source
AI summary
A communication control method performed by a first communication apparatus configured to perform wireless communication with a second communication apparatus in a mobile communication system using a machine learning technology, the communication control method including learning by performing model learning through which a learned model is derived by using learning data including a reception signal from the second communication apparatus, and controlling transmission and/or reception of control data related to the model learning to and from the second communication apparatus.