UE Power Saving via Data Throughput Pattern Prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In wireless communications, especially under 3GPP specifications for 5G New Radio (NR), user equipment (UE) experiences unnecessary power consumption when staying in a RRC connected state without data transmission, due to the RRC inactivity timer expiration.

Innovation Solution

The UE determines the probability of a continuous duration of no uplink and downlink data exceeding the RRC inactivity timer duration plus a threshold, and triggers an early RRC connection release if this probability is higher than a set threshold, thereby reducing power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If the UE stays in RRC connected state for the predefined period to wait for RRC inactivity timer expiry, then the network can trigger connection release, but the UE experiences unnecessary power consumption during this waiting period

Engineering Contradiction:

Improvepower consumptionVSAvoidconnection release timing accuracy

Core Design Contradiction:

Loss of energyVSLoss of time

Solution Approach 1:

The UE performs preliminary actions by predicting future data arrival patterns using machine learning models before the RRC inactivity timer expires. The system analyzes historical data throughput patterns and screen-on/off status to forecast whether data will arrive within the remaining timer period. If prediction indicates no data will arrive, the UE triggers early RRC connection release, avoiding the need to wait for timer expiry and thereby reducing power consumption while maintaining accurate connection release timing.

Inventive Principle:

Principle #10Preliminary action

2Loss of energy

If the UE triggers early RRC connection release based on data throughput pattern prediction, then power consumption is reduced, but the prediction accuracy must be sufficient to avoid premature release

Engineering Contradiction:

Improvepower consumptionVSAvoidprediction accuracy

Core Design Contradiction:

Loss of energyVSReliability

Solution Approach 1:

The system implements feedback mechanisms where the UE continuously monitors actual data arrival patterns against predicted patterns. The machine learning models are trained using historical data throughput patterns and screen-on/off status, and the prediction accuracy is continuously improved through feedback from actual network conditions. This feedback loop ensures that early RRC connection release decisions are made with sufficient prediction accuracy, preventing premature release while maximizing power savings.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system dynamically adjusts prediction parameters and decision thresholds based on changing network conditions and UE behavior patterns. By modifying parameters such as prediction confidence thresholds, time window durations, and feature weighting in the machine learning model, the system adapts to different scenarios (e.g., screen-on vs. screen-off states) to maintain reliable prediction accuracy while optimizing power consumption reductions.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20250048493A1Power Saving By Data Throughput Pattern Prediction In Wireless Communications

Publication Date: 2025.02.06 MEDIATEK SINGAPORE PTE LTD

AI summary

Techniques pertaining to power saving by data throughput pattern prediction in wireless communications are described. A user equipment (UE) determines whether a probability of a first value being greater than a second value is higher than a threshold. The UE triggers a radio resource control (RRC) connection release with a network responsive to the probability being higher than the threshold. The first value represents a succeeding continuous duration of no uplink (UL) and downlink (DL) data. The second value represents an RRC inactivity timer duration plus a threshold duration.