Federated Learning Circle-of-Trust for Secure UE Model Sharing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing federated learning architectures fail to address privacy concerns and data security in communication systems, particularly in networks with diverse UEs, leading to reduced AI/ML model accuracy and limited utility due to the lack of mechanisms for identifying and securely transporting data between UEs and network entities.

Innovation Solution

Implementing a primary UE that establishes a circle-of-trust with secondary UEs, using cryptographic techniques and security protocols to secure AI/ML models and data, ensuring privacy by encrypting and anonymizing sensitive information before sharing with a network entity, while allowing for collaborative learning.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If federated learning is implemented without secure data transport mechanisms, then collaborative learning can proceed, but privacy concerns arise and data security is compromised

Engineering Contradiction:

Improvecollaborative learning capabilityVSAvoiddata security

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces a primary UE as an intermediary that establishes a circle-of-trust with secondary UEs. This intermediary manages secure data transport by encrypting data before transmission to the network entity, thus enabling collaborative learning while maintaining data security and privacy.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements preliminary cryptographic operations (encryption and anonymization) on data before it leaves the UEs. By performing these security measures in advance, the system enables collaborative learning to proceed while ensuring data security and privacy are maintained throughout the federated learning process.

Inventive Principle:

Principle #10Preliminary action

2Reliability

If data is encrypted and anonymized before sharing, then privacy is maintained, but data processing complexity increases

Engineering Contradiction:

Improveprivacy protectionVSAvoiddata processing complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent segments the federated learning system into distinct roles (primary UE and secondary UEs) with specific responsibilities. The primary UE handles the complex cryptographic operations of encryption and anonymization, while secondary UEs focus on providing training data and receiving model updates, thus distributing processing complexity across multiple entities.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system implements self-service mechanisms where UEs automatically perform encryption and anonymization of their own data before sharing. This automated approach reduces manual intervention and simplifies the overall process despite the cryptographic complexity involved.

Inventive Principle:

Principle #25Self-service

3Measurement precision

If AI/ML models are tailored to individual UE characteristics, then model accuracy improves, but data sharing requirements increase privacy risks

Engineering Contradiction:

Improvemodel accuracyVSAvoidprivacy risks

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The primary UE acts as an intermediary that enables individualized model training by securely managing data from multiple secondary UEs. It encrypts and anonymizes data before transmission, allowing the network entity to train accurate individualized models while the UEs maintain privacy through the cryptographic protections provided by the intermediary.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent implements local quality by allowing each UE to have its own tailored AI/ML model trained on its specific characteristics and data patterns. Simultaneously, it applies uniform cryptographic protection (encryption and anonymization) to all data shared with the network entity, thus achieving both model personalization and privacy protection.

Inventive Principle:

Principle #3Local quality

4Reliability

If secure cryptographic protocols are implemented for data transport, then data security is enhanced, but communication overhead increases

Engineering Contradiction:

Improvedata securityVSAvoidcommunication overhead

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent segments cryptographic operations into two main components: encryption of training data before transmission, and anonymization of identifiers. This segmentation allows the system to apply cryptographic protections only where necessary (on data leaving UEs) rather than throughout the entire federated learning process, thus reducing overall communication overhead while maintaining security.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250344064A1Systems and methods for secure transport and processing of data in federated learning

Publication Date: 2025.11.06 VERIZON PATENT & LICENSING INC

AI summary

In some implementations, a first UE may identify a configuration that includes artificial intelligence or machine learning (AI/ML) model parameters to be used and shared for federated learning. The first UE may generate an AI/ML model based on the configuration, wherein the AI/ML model is based on an anonymization and encryption of one or more information elements (IEs) using policy information. The first UE may secure the AI/ML model. The first UE may establish a circle-of-trust to include the first UE and a second UE. The first UE may transmit the AI/ML model to the second UE based on the second UE being included in the circle-of-trust.