AI Model Identifier Registration for Cross-Network Invocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In AI application scenarios, terminal devices and access network devices lack a unified understanding of AI models trained by the access network device, affecting their use effectiveness.
Innovation Solution
A communication method and apparatus that enables the registration of AI models by obtaining and sending request information with identifiers, allowing devices within the same operator's coverage to have a unified understanding of the model, facilitating seamless invocation and management across network devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the access network device trains and transmits the AI model to the terminal device, then the AI model can be deployed for use, but the terminal device and access network device do not have a unified understanding of the AI model, affecting use effectiveness
Solution Approach 1:
The patent applies preliminary action by having the terminal device send model identifier information to the access network device before the AI model is actually used. This pre-registration of model identifiers ensures that both devices have a unified understanding of the AI model from the outset, eliminating the information asymmetry problem that would otherwise affect use effectiveness.
2Adaptability or versatility
If the AI model is trained by the access network device, then the model can be customized for network-specific requirements, but devices in the application scenario do not have a unified understanding of the model
Solution Approach 1:
The patent implements feedback by establishing a communication mechanism where the terminal device reports model identifier information back to the access network device. This feedback loop ensures that the access network device maintains awareness of the AI models deployed at terminal devices, enabling unified understanding while preserving the adaptability of locally trained models.
3Ease of operation
If the terminal device invokes the AI model using the first identifier after moving to another access network device's coverage, then seamless model invocation is enabled, but the system complexity increases
Solution Approach 1:
The patent applies universality by designing a model identifier registration system that works across different access network devices. The same model identifier and registration mechanism used within a single access network device's coverage can be universally applied when the terminal device moves to another access network device's coverage, enabling seamless model invocation without increasing system complexity.
Data Source
Figure 1~3
Figure 4
Figure 5
AI summary
Embodiments of this application provide a communication method and apparatus. The method is applied to registration of an AI model. The method includes: a first network device obtains a first model through training; and the first network device sends first request information to request to register the first model with a second network device, where the first request information includes a first identifier corresponding to the first model. The first model is registered, so that devices in an application scenario of the first model can have a unified understanding of the first model, to improve application effect of the first model.