Policy Function Network Element for 5G Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 5G mobile communications systems, there is a lack of a control mechanism to convert data information into policies using machine learning technology, particularly in determining the specific function division between network elements and implementing machine learning to output policies in wireless networks.
Innovation Solution
A policy-driven method and apparatus that utilizes a collaboration of function network elements to convert data information into policies by obtaining, identifying, and executing policies using machine learning, allowing for the reuse of models and efficient data transmission between network elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional BBU and RRU network elements are used with fixed function division, then network architecture is simple, but adaptability to different service requirements and transmission needs is poor
Solution Approach 1:
The patent segments the base station into multiple independent network elements (CU, DU, RRU) with separable functions. Each element can be independently deployed and configured, allowing flexible adaptation to different service requirements while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent implements dynamic function division where network elements can be flexibly configured and reconfigured based on service requirements. The separation of control plane and user plane functions allows dynamic adaptation without requiring complete network redesign, balancing adaptability with architectural simplicity.
2Measurement precision
If machine learning models are coupled together to convert data information into policies, then policy accuracy is improved, but model reuse efficiency decreases and system complexity increases
Solution Approach 1:
The patent extracts the policy generation function from the data processing flow, creating a separate policy management module. This allows multiple data service function network elements to independently process data and generate results without requiring complex inter-model coupling, improving reuse efficiency while maintaining policy accuracy through centralized policy management.
Solution Approach 2:
The patent introduces a policy management module as an intermediary between data processing and policy execution. This mediator receives data from multiple sources, applies appropriate policies based on data characteristics, and generates accurate results without requiring direct coupling between models, thereby improving both accuracy and reuse efficiency.
3Measurement precision
If multiple network elements collaborate to process data information, then policy control precision is improved, but data transmission overhead and processing time increase
Solution Approach 1:
The patent implements preliminary data processing at the data service function network element level, where data is pre-processed and filtered before being transmitted to the policy management module. This preliminary action reduces the complexity of subsequent policy processing, maintaining high precision while minimizing additional processing time introduced by multi-element collaboration.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of this application disclose a policy-driven method and an apparatus. Through collaboration of a plurality of function network elements, data information in a wireless network may be converted into a policy by using a machine learning technology. The method includes: obtaining, by a policy function network element, at least one policy from a modeling function network element; receiving, a first policy identifier triggered by a first model from a prediction function network element, where the first policy identifier is used to identify a first policy; determining, by the policy function network element, the first policy in the at least one policy based on the first policy identifier, where the first policy includes action information corresponding to the first action; executing, by the policy function network element, the first action in the first policy; and sending, by the policy function network element, an execution result of the first action to a data service function network element or the prediction function network element, where the execution result of the first action is used as input data of a second model.