AI Model Matching and Pushing for 5G Network Nodes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

There is a lack of a unified technical solution for selecting and managing a large number of artificial intelligence models in telecommunication networks, particularly in the context of 5G networks, which are complex due to the introduction of SDN and NFV, leading to challenges in model deployment and optimization.

Innovation Solution

A method and device for model pushing and requesting that utilize an orchestrator to generate model matching instructions, search for appropriate models, and deploy them to destination nodes, with mechanisms for updating and optimizing models to ensure accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a large number of artificial intelligence models are introduced to meet diverse analysis requirements in 5G networks, then the model selection capability and analysis coverage are improved, but the complexity of model management and deployment increases significantly

Engineering Contradiction:
Improvemodel selection capabilityVSAvoidmodel management complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent introduces a model management platform as an intermediary between model providers and network operators. This platform includes a model repository for centralized storage, a model matching module for automatic selection based on analysis requirements, and a model deployment module for automated deployment. The intermediary abstracts the complexity of managing numerous models, allowing operators to select and deploy appropriate models without directly dealing with the underlying complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The model management platform is designed as a universal system that can handle multiple types of artificial intelligence models (machine learning, deep learning, reinforcement learning) and serve multiple analysis requirements (traffic prediction, fault detection, resource optimization). The platform provides unified interfaces and standardized processes for model registration, storage, selection, deployment, and updates, making it adaptable to diverse model types and application scenarios while maintaining consistent management procedures.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Measurement precision

If manual model selection and deployment processes are used, then model deployment accuracy can be controlled, but the time consumption and operational efficiency are significantly reduced

Engineering Contradiction:
Improvemodel deployment accuracyVSAvoidmodel deployment time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The model management platform implements automated self-service mechanisms where the system automatically selects appropriate models based on analysis requirements, retrieves them from the repository, and deploys them to target nodes without manual intervention. The model matching module automatically matches models to requirements by comparing model capabilities with analysis needs, and the deployment module automatically provisions and configures models on target infrastructure, significantly reducing deployment time while maintaining accuracy through automated validation processes.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system incorporates feedback mechanisms where deployment results and model performance data are collected and fed back to the model management platform. This feedback is used to validate deployment accuracy, identify successful model-analys is matches, and continuously improve the model matching algorithm. The feedback loop ensures that automated processes maintain high accuracy by learning from actual deployment outcomes and adjusting selection criteria accordingly.

Inventive Principle:
Principle #23Feedback

3Extent of automation

If existing models are stored in distributed locations, then system autonomy is improved, but the difficulty of model searching and retrieval increases

Engineering Contradiction:
Improvesystem autonomyVSAvoidmodel search difficulty
Core Design Contradiction:
Extent of automationVSDifficulty of detecting and measuring

Solution Approach 1:

The patent merges distributed model storage into a centralized model repository while maintaining the autonomous operation of distributed model deployment. The centralized repository consolidates all artificial intelligence models in a standardized format with metadata describing model capabilities, requirements, and characteristics. This consolidation enables efficient searching, filtering, and retrieval of models based on analysis requirements, while the model deployment module maintains autonomous deployment capabilities to distributed target nodes, combining the benefits of centralized management and distributed execution.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP3958606B1Model pushing method and device, model requesting method and device, storage medium and electronic device
Publication Date: 2025.07.30 ZTE CORP
  • EP3958606B1 patent drawingFigure 1~2
  • EP3958606B1 patent drawingFigure 3~5
  • EP3958606B1 patent drawingFigure 6

AI summary

The present disclosure provides a model pushing method, a model requesting method, a model pushing device, a model requesting device, a storage medium and an electronic device. The model pushing method includes: receiving a model matching instruction sent by an orchestrator, with the model matching instruction generated based on an analysis requirement; searching for a model corresponding to the model matching instruction; and pushing, in a case where the model is found, the found model to a destination node requiring the model.