ML Model Access Tokens for Secure Network Retrieval

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current communication systems face inefficiencies in network data analytics services, particularly in the retrieval of machine learning models, with issues such as unauthorized access, lack of defined authorization schemas, and challenges in transferring large model sizes over network interfaces.

Innovation Solution

Implementing an authorization schema for machine learning model retrieval using OAuth 2.0 access tokens enhanced with model identifiers, enabling authorized access and transfer of machine learning models through Analytics Data Repository Functions and Network Function producers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional network data analytics services are used for machine learning model retrieval, then service availability is maintained, but security authorization is insufficient and unauthorized access cannot be prevented

Engineering Contradiction:
Improvesecurity authorizationVSAvoidauthorization schema complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces an OAuth 2.0 authorization server as an intermediary component that mediates between machine learning model producers and consumers. This server issues access tokens that enable secure, standardized authorization without requiring direct complex trust relationships between all network entities, thus improving security while managing complexity through a dedicated intermediary service.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent changes the authorization parameter system by implementing OAuth 2.0 access tokens with specific claims (model identifier, consumer identifier, permissions) that dynamically encode authorization information. This parameter-based approach replaces traditional static authorization methods, enabling flexible security control while maintaining system manageability through standardized token structures.

Inventive Principle:
Principle #35Parameter changes

2Ease of operation

If machine learning models are transferred over network interfaces, then model accessibility is improved, but bandwidth consumption increases due to large model sizes

Engineering Contradiction:
Improvemodel accessibilityVSAvoidbandwidth consumption
Core Design Contradiction:
Ease of operationVSLoss of energy

Solution Approach 1:

The patent segments the machine learning model transfer process into distinct phases: authorization token issuance, model retrieval requests with token validation, and actual model data transfer. This segmentation allows for optimized transfer protocols where only authorized consumers can initiate transfers, and the authorization metadata is separated from the model payload, reducing redundant signaling and optimizing bandwidth utilization.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements preliminary authorization actions through OAuth 2.0 token issuance before actual model transfer occurs. Consumers must obtain access tokens with valid authorization claims beforehand, which are then validated during the transfer process. This preliminary authorization step prevents unauthorized transfer attempts, reducing wasted bandwidth on unsuccessful transfers and enabling efficient resource allocation.

Inventive Principle:
Principle #10Preliminary action

3Ease of operation

If network data analytics functions are deployed to provide analytics data, then data consumer service is improved, but service efficiency decreases and data consumers may not receive data for some analytics identifications

Engineering Contradiction:
Improvedata consumer serviceVSAvoidservice efficiency
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The patent implements feedback mechanisms where the authorization server and model repository function validate access tokens and provide authorization status information back to consumers. This feedback loop ensures that consumers receive clear information about their authorized access rights, enabling them to efficiently retrieve appropriate models without unnecessary requests or rejections, thus improving service efficiency while maintaining ease of operation.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4270872B1Authorized machine learning model retrieval for a communications network
Publication Date: 2026.04.22 NOKIA TECHNOLOGIES OY
  • EP4270872B1 patent drawingFigure 1
  • EP4270872B1 patent drawingFigure 2
  • EP4270872B1 patent drawingFigure 3

AI summary

Methods, systems, apparatuses, and computer program products are provided for authorized machine learning model retrieval for a communications network. In this regard, an access token request for one or more machine learning models related to a communications network is received from a network function service consumer (NFc). The access token request includes information to identify the one or more machine learning models. The NFc is then authorized with respect to the one or more machine learning models based on the information included in the access token request. Additionally, enhanced an access token for retrieving the one or more machine learning models is provided to the NFc based on valid authorization of the NFc with respect to the one or more machine learning models.