AI Model Input Mapping for Flexible Network Data Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile communication systems lack flexibility in selecting input data for AI models, limiting the effectiveness of AI-based applications such as CSI feedback compression, beam management, and load balancing.
Innovation Solution
A method and apparatus that enable a device to obtain first and second information for determining target data and its mapping relationship with model inputs, allowing for flexible selection of input data based on these mappings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If predefined procedures are used to obtain model input data, then the system is simple to implement, but the flexibility in selecting input data is poor
Solution Approach 1:
The patent implements dynamic selection of input data for AI models by allowing the network device to flexibly determine which data to acquire based on current network conditions and model requirements, rather than using fixed predefined procedures. This enables the system to adapt to different scenarios while maintaining manageable complexity through structured data acquisition methods.
2Measurement precision
If more input data is collected for AI model inference, then the inference accuracy is improved, but the data acquisition time and resource consumption increase
Solution Approach 1:
The patent applies partial action by selectively acquiring only the necessary input data required for AI model inference rather than collecting all available data. The network device determines the specific data needed based on the model requirements and current network state, achieving sufficient inference accuracy while minimizing data acquisition time and resource consumption.
Solution Approach 2:
The patent changes the parameter of data selection from fixed to variable, allowing the network device to dynamically adjust which input data to acquire based on inference requirements. This enables optimization of the balance between inference accuracy and acquisition time by adapting the data set to specific operational conditions.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
This application discloses a model processing method and apparatus, a terminal, and a network side device, and belongs to the field of communication technologies. The model processing method in embodiments of this application includes: obtaining, by a first device, first information and second information, where the first information is for determining target data for target model inference, and the second information is for indicating a mapping relationship between the target data for the target model inference and an input of a target model; performing, by the first device, data acquisition based on the first information, to obtain the target data; and determining, by the first device, input data of the target model based on the second information and the target data.