Predictive Handover Measurement Logging for Lower UE Signaling

Solution Overview

Problem

The high frequency of measurement reporting for machine learning-based predictive handover in wireless cellular systems leads to significant consumption of UE resources and signaling overhead, necessitating an efficient data collection mechanism.

Innovation Solution

A handover-events logging unit (HELU) in the UE manages data collection for predictive ML-based handover by implementing optimized logging and reporting conditions, such as Start-log, Stop-log, and Report-log events, to minimize unnecessary data transfer and enhance training data relevance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If measurement reporting frequency is increased to improve ML-based predictive handover accuracy, then handover prediction accuracy is improved, but UE resource consumption and signaling overhead increase significantly

Engineering Contradiction:

Improvehandover prediction accuracyVSAvoidUE resource consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The network performs preliminary actions by configuring the UE with specific measurement reporting criteria and thresholds before handover events occur. The UE logs measurement data according to pre-configured conditions (Start-log, Stop-log, Report-log events) rather than continuously reporting, reducing resource consumption while maintaining prediction accuracy through targeted data collection at critical moments.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention changes the parameter of measurement reporting from continuous high-frequency reporting to event-driven selective reporting. By modifying the reporting mechanism to trigger only at specific events (Start-log, Stop-log, Report-log), the system maintains measurement precision for handover prediction while significantly reducing UE resource consumption and signaling overhead.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If measurement reporting frequency is increased to improve ML-based predictive handover accuracy, then handover prediction accuracy is improved, but signaling overhead increases significantly

Engineering Contradiction:

Improvehandover prediction accuracyVSAvoidsignaling overhead

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The network performs preliminary actions by configuring the UE with specific measurement reporting criteria and thresholds before handover events occur. The UE logs measurement data according to pre-configured conditions (Start-log, Stop-log, Report-log events) rather than continuously reporting, reducing resource consumption while maintaining prediction accuracy through targeted data collection at critical moments.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention changes the parameter of measurement reporting from continuous high-frequency reporting to event-driven selective reporting. By modifying the reporting mechanism to trigger only at specific events (Start-log, Stop-log, Report-log), the system maintains measurement precision for handover prediction while significantly reducing UE resource consumption and signaling overhead.

Inventive Principle:

Principle #35Parameter changes

3Quantity of substance

If continuous measurement data is collected to train ML models, then training data availability is improved, but data relevance and quality decrease due to unnecessary data transfer

Engineering Contradiction:

Improvetraining data availabilityVSAvoiddata relevance

Core Design Contradiction:

Quantity of substanceVSLoss of information

Solution Approach 1:

The invention extracts only the necessary measurement data at critical moments rather than continuously collecting all data. By using event-driven logging (Start-log when entering handover region, Stop-log when leaving, Report-log for actual handovers), the system extracts relevant training data while filtering out unnecessary information, improving data relevance and quality for ML model training.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system applies local quality by collecting measurement data selectively at specific locations and moments (handover regions, event triggers) rather than uniformly throughout. This ensures that training data is both sufficient in quantity and high in quality, as only measurement data relevant to handover events is collected and transmitted.

Inventive Principle:

Principle #3Local quality

Data Source

PatentEP3962153B1Signalling data for training machine learning based predictive handover

Publication Date: 2025.12.10 NOKIA TECHNOLOGIES OY

Figure 1
Figure 2
Figure 3

AI summary

There is provided an apparatus for measurement collection for training a machine learning algorithm for predictive handover, the apparatus comprising means for: receiving a configuration message from a network, the configuration message comprising at least information on logging radio signal measurements data that is correlated to handover events and a reporting condition for reporting at least part of the radio signal measurements to be logged by the apparatus; in response to fulfilling a starting condition, starting logging of measurements at least partly based on the configuration message, wherein the starting condition is stored in the apparatus and/or comprised in the configuration message; in response to fulfilling a stopping condition, checking whether the reporting condition is fulfilled, wherein the stopping condition is stored in the apparatus and/or comprised in the configuration message; discarding the logged measurements if the reporting condition is not fulfilled; or reporting to the network at least part of the logged measurements if the reporting condition is fulfilled.