Context-Based CNF and ML Model Selection for Network Automation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing communication network management systems struggle to efficiently select and deploy cognitive network functions (CNFs) and machine learning (ML) models that match the network's expected or detected context, especially when multiple vendors and models are involved, lacking mechanisms for context-based selection and standardized information exchange.
Innovation Solution
A cognitive network function and model repository function (CoMReF) and selection function (CoMSeF) are introduced, along with a structured interface (CoMSI) and information model to enable operators to request and select context-specific CNFs and ML models, using standardized metadata to match network requirements with available models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple cognitive network functions and machine learning models from different vendors are deployed, then the versatility and functionality of the network are improved, but the complexity of selecting and managing the appropriate models increases
Solution Approach 1:
The patent introduces an intermediary system comprising a model repository function and selection function that acts as a mediator between multiple ML model vendors and the network operator. This intermediary standardizes the information exchange format, centralizes model storage, and automates the selection process based on network context, thereby resolving the complexity issue while maintaining versatility.
Solution Approach 2:
The patent creates a universal interface and standardized information model that can handle multiple ML models from different vendors through a single standardized channel. The selection function serves multiple purposes: storing models, evaluating them against network context, and selecting the optimal model, thereby reducing overall system complexity through multi-functionality.
2Measurement precision
If context-based selection mechanisms are implemented, then the precision of model matching is improved, but the information exchange requirements and system complexity increase
Solution Approach 1:
The patent transforms the model selection problem into a parameter-matching problem by defining specific context parameters (network state, performance metrics, operational conditions) and model parameters (input data characteristics, expected output, performance requirements). The selection function compares these parameters systematically to achieve precise matching while standardizing the information exchange format to minimize data loss.
Solution Approach 2:
The patent implements preliminary action by pre-defining the information model structure, pre-storing ML models with their associated context requirements in the repository, and pre-establishing the selection criteria based on network context parameters. This preliminary preparation reduces the information exchange burden during actual model selection operations.
3Productivity
If automated model selection and deployment is implemented, then the productivity of network management is improved, but the extent of automation required increases system complexity
Solution Approach 1:
The patent implements self-service by enabling the selection function to autonomously evaluate network context, query the model repository, select the appropriate ML model, and deploy it without human intervention. The system monitors network conditions and automatically triggers model reselection when context changes, achieving high productivity through self-service automation while keeping the automation logic centralized and manageable.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The disclosure relates to an apparatus comprising means for: storing (900) information relating to a plurality of cognitive network functions and/or machine learning models deployable on a network, each cognitive network function being associated with one or more machine learning models, each machine learning model being trained with one or more training contexts; receiving (902) a first request for a cognitive network function and/or a machine learning model fulfilling a requirement or for information relating to a cognitive network function and/or a machine learning model fulfilling a requirement; processing (904) the first request to identify a cognitive network function and/or a machine learning model fulfilling the requirement; and sending (906) a response to the first request based on the processing.