Kernel Event Log Analysis for Power Consumption Estimation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing power management techniques for electronic devices are inefficient in accurately estimating power consumption by software applications, leading to suboptimal battery life and user experience, especially in portable devices where energy efficiency is crucial.

Innovation Solution

A method using a training module and current usage module to analyze log data and generate a current consumption model through machine learning, allowing for real-time estimation of power usage by mapping kernel event logs to current consumption values, enabling accurate prediction of power consumption across various user activities.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional power measurement methods are used, then power consumption data can be obtained, but the process is labor-intensive and costly

Engineering Contradiction:
Improvepower consumption estimation accuracyVSAvoidmeasurement process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent creates a computational model that replicates the function of physical power measurement equipment. Instead of using actual power meters and measurement apparatus, the system uses machine learning models trained on kernel event logs to predict power consumption, thereby copying the measurement function through software rather than hardware.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces the mechanical/physical measurement system with an information-processing system. Rather than physically measuring power consumption through electrical instruments, the system uses software-based machine learning models that process kernel event logs to estimate power consumption, substituting physical measurement with computational prediction.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If detailed power measurement is performed, then accurate power consumption data is obtained, but time and resources are consumed

Engineering Contradiction:
Improvepower consumption measurement accuracyVSAvoidmeasurement time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary training of machine learning models using power consumption data collected during typical device operation. The models are pre-trained on kernel event logs that correlate with power consumption patterns, so that during actual use, the system can quickly predict power consumption without performing time-consuming real-time measurements.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates a computational replica of the power measurement process through machine learning models. These models capture the relationship between kernel events and power consumption, allowing the system to estimate power usage instantly by processing log data rather than performing actual physical measurements.

Inventive Principle:
Principle #26Copying

3Productivity

If machine learning models are used for power estimation, then labor and costs are reduced, but model accuracy must be maintained

Engineering Contradiction:
Improvepower estimation efficiencyVSAvoidpower consumption prediction accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements a feedback mechanism where the machine learning model's predictions are continuously refined. The system trains models on historical kernel event logs with associated power consumption data, uses the trained models to predict power consumption, and can further improve accuracy by incorporating actual measured data back into the training set for iterative refinement.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent optimizes model accuracy by adjusting various parameters including the selection of kernel events included in the training data, the choice of machine learning algorithms, feature engineering parameters, and model hyperparameters. This allows the system to achieve high prediction accuracy while maintaining computational efficiency.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2772862B1Electrical current estimation for electronic devices
Publication Date: 2017.12.20 BLACKBERRY LTD
  • EP2772862B1 patent drawingFigure 1
  • EP2772862B1 patent drawingFigure 2~3
  • EP2772862B1 patent drawingFigure 4

AI summary

Various embodiments are described herein for a system and method for estimating current consumption for an electronic device by obtaining log data comprising a record of at least some activities of the electronic device during a selected time period, parsing the log data into a plurality of component digests, estimating current consumption values for the component digests based on component signatures of the component digests, a current consumption model and a machine learning technique; and processing the estimated current consumption values to estimate current consumption data for the electronic device.