Neural Network Model Adjustment for Communication Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AI transceivers can only adapt to specific communication scenarios, requiring retraining when scenarios change, which is complex and increases signaling overheads in general-purpose communication systems.
Innovation Solution
A method to determine and deploy a neural network model for communication between transmit and receive ends with low complexity and signaling overheads by adjusting pre-trained neural network models based on communication resource information and channel state information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a current AI transceiver is used to adapt to a specific communication scenario, then communication performance in that scenario is improved, but device complexity increases when scenarios change requiring retraining
Solution Approach 1:
The patent applies preliminary action by pre-training neural network models in advance for different communication scenarios. Instead of retraining models when scenarios change, the system prepares multiple pre-trained models beforehand, allowing quick selection and switching based on current communication conditions, thereby avoiding complex retraining operations.
Solution Approach 2:
The patent utilizes parameter changes by adjusting model selection based on communication scenario parameters. When communication conditions change, the system changes the parameter of which pre-trained model is selected or deployed, rather than modifying the model itself through retraining, thus reducing complexity while maintaining adaptability.
2Measurement precision
If neural network models are retrained for different communication scenarios, then model accuracy for specific scenarios is improved, but signaling overheads increase
Solution Approach 1:
The patent applies copying by creating multiple pre-trained model copies for different communication scenarios. Each scenario has its own pre-trained model copy that can be directly deployed without retraining. This eliminates the need for extensive signaling to coordinate retraining between devices, reducing signaling overheads while maintaining high accuracy for each specific scenario.
3Reliability
If pre-trained neural network models are adjusted based on communication resource information and channel state information, then communication quality is improved, but model determination complexity increases
Solution Approach 1:
The patent applies dynamics by making the model selection and adjustment process adaptive to changing communication conditions. The system dynamically selects and adjusts pre-trained models based on real-time communication resource information and channel state information, allowing the model determination process to flexibly respond to varying conditions without excessive complexity.
Data Source
AI summary
This application provides a communication method and a related apparatus. In this method, a first neural network model may be determined from one or more pre-trained neural network models, the first neural network model is adjusted based on communication resource information and/or channel state information for communication between a first device and a second device, and an adjusted first neural network model is sent, or a submodel in an adjusted first neural network model is sent, where the adjusted first neural network model is used for the communication between the first device and the second device. It can be learned that the neural network model used for the communication between the first device and the second device is obtained through adjustment based on the pre-trained neural network model that has been trained.