5G Slice Selection Function Self-Learning for Decision Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 5G networks, the slice selection function faces challenges in making accurate slice reselection decisions due to insufficient experiences and obsolete context information, leading to potential wrong decisions and suboptimal quality of experience (QoE).
Innovation Solution
A method where the slice selection function learns slice reselection decisions by determining if enough self-operation cases exist, requesting evaluation from a management function, and receiving evaluation results to update experiences, while the management plane function evaluates and manages slice reselection experiences to correct wrong decisions and adapt to context changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the slice selection function makes reselection decisions based on limited self-operation cases, then the decision-making process is fast and autonomous, but the accuracy of decisions deteriorates due to insufficient experiences
Solution Approach 1:
The patent implements a feedback mechanism where the management function evaluates the slice selection function's decisions and provides correction information. The slice selection function learns from these evaluation results, continuously improving decision accuracy while maintaining autonomous operation. This resolves the contradiction by enabling accurate decisions through feedback-based learning without sacrificing decision-making speed.
Solution Approach 2:
The slice selection function performs self-learning by automatically incorporating evaluation results from the management function into its decision-making process. This self-service mechanism allows the system to improve its own accuracy over time without external intervention, maintaining fast autonomous operation while progressively enhancing decision quality.
2Measurement precision
If the slice selection function collects and evaluates more self-operation cases, then the decision accuracy improves, but the system complexity and evaluation overhead increase
Solution Approach 1:
The patent divides the evaluation system into two segments: the slice selection function that makes rapid decisions based on current experiences, and the management function that periodically evaluates accumulated cases. This segmentation allows accurate decisions through comprehensive evaluation while maintaining system simplicity by separating real-time operation from batch evaluation.
Solution Approach 2:
The management function performs evaluation periodically rather than continuously, assessing accumulated self-operation cases at intervals. This periodic action reduces evaluation overhead and system complexity while still improving decision accuracy through regular learning updates, avoiding the need for complex real-time evaluation mechanisms.
3Productivity
If the slice selection function uses obsolete context information, then the system operation is simple and fast, but the quality of experience deteriorates due to wrong decisions
Solution Approach 1:
The management function provides feedback containing corrections based on obsolete context information, enabling the slice selection function to learn and update its knowledge base. This feedback mechanism maintains fast operation by using simple current rules while improving reliability over time through periodic corrections of obsolete information.
Solution Approach 2:
The management function performs preliminary evaluation of context information and prepares correction data in advance. This preliminary action ensures that when the slice selection function operates with simple fast rules, it does so with up-to-date contextual understanding, improving quality of experience without sacrificing operational speed.
Data Source
AI summary
A method includes learning, by at least one device associated with a slice selection function, a slice reselection decision for at least one self-operation case. The method includes determining, by the at least one device associated with the slice selection function, whether an evaluation condition for at least one new self-operation case has been met. The at least one new self-operation case includes the at least one self-operation case. The method also includes requesting, at least one device associated with a management function, evaluate the at least one new self-operation case in response to determining that the evaluation condition has been met, receiving evaluation results from the at least one device associated with the management function, and learning the evaluation results.