UE Training Data Prioritization for Handover Memory Constraints

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The limited memory of user equipment (UE) in wireless communication systems can lead to training data override issues during handovers, degrading the accuracy of AI/ML models due to insufficient memory causing unintended data modification or loss.

Innovation Solution

Implement prioritization for training data collection by providing prioritization information based on vendor, AI/ML use case, network identifier, and carrier frequency, ensuring that training data collection aligns with available memory and prioritization levels, thereby preventing data override and maintaining model accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If training data collection continues without prioritization during handover, then more training data can be collected, but memory overflow occurs causing data override and loss

Engineering Contradiction:

Improvetraining data volumeVSAvoiddata integrity

Core Design Contradiction:

Quantity of substanceVSReliability

Solution Approach 1:

The patent applies local quality by assigning different priority levels to different training data based on their importance to AI/ML model performance. High-priority data is preserved while low-priority data is discarded when memory is full, ensuring data integrity for critical information while allowing continuous collection of less critical data.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent changes the parameter of data priority classification, introducing a priority level parameter that categorizes training data into high, medium, and low priorities. This parameter change enables the system to make informed decisions about which data to retain or discard during memory constraints.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If UE memory is expanded to store all training data, then data override is prevented, but device complexity and cost increase

Engineering Contradiction:

Improvedata integrityVSAvoidmemory capacity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements discarding and recovering by systematically discarding low-priority training data when memory is full, while preserving high-priority data. This approach maintains data integrity for critical information without requiring expanded memory capacity, avoiding increased device complexity.

Inventive Principle:

Principle #34Discarding and recovering

3Reliability

If prioritization is implemented for training data collection, then data override protection is improved, but signaling overhead increases

Engineering Contradiction:

Improvedata override protectionVSAvoidsignaling overhead

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The patent applies partial action by transmitting prioritization information selectively - only when necessary for data override protection - rather than continuously. This reduces signaling overhead while maintaining effective data priority management and override protection.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20260082185A1Prioritization for training data collection

Publication Date: 2026.03.19 QUALCOMM INC

AI summary

Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may receive, from a first network node, a training data collection configuration that indicates prioritization information, wherein a training data override protection is based at least in part on the prioritization information. The UE may collect training data based at least in part on the training data collection configuration and the prioritization information. The UE may transmit, to the first network node, the training data. Numerous other aspects are described.