Predictive Power Management for Computing Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing energy consumption techniques for computing devices are backward-looking and lack predictive value, failing to estimate energy consumption for power management settings that have not been employed, limiting their ability to optimize energy use proactively.
Innovation Solution
The implementation of predictive power management tools that measure resource usage under current power policies and estimate energy consumption for different policies without actually running them, using power models and machine learning to simulate and recommend optimal power settings based on user preferences and community data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing energy consumption techniques measure past usage and extrapolate forward, then energy consumption for current settings can be estimated, but predictive value for other power management settings cannot be provided
Solution Approach 1:
The system performs preliminary measurements of energy consumption across multiple power management settings during an initial characterization phase. These pre-collected data points enable the predictive model to estimate energy consumption for any setting without actually implementing it, resolving the contradiction between measurement precision and predictive versatility
Solution Approach 2:
The system creates a virtual model (copy) of the computing device's power consumption characteristics by measuring actual energy usage under various settings. This copied model can then predict energy consumption for untested settings without requiring physical implementation, providing predictive value while maintaining measurement accuracy
2Loss of energy
If power management settings are changed to optimize energy consumption, then energy conservation is improved, but the time required to assess the impact of setting changes increases
Solution Approach 1:
The system performs preliminary characterization of energy consumption across all power management settings before the user needs to make decisions. This pre-computed data enables instant assessment of energy conservation benefits without requiring the user to wait for actual usage periods, resolving the time-energy assessment contradiction
Solution Approach 2:
The system provides immediate feedback to users about the predicted energy consumption impact of different power management settings based on the pre-built model. This instantaneous feedback enables users to make informed decisions about energy optimization without experiencing delays, balancing energy conservation goals with user convenience
Data Source
AI summary
The described implementations relate to predictive computing device energy management. One implementation measures resource usage of a computing device that employs a power policy. This implementation also estimates resource usage of the computing device having at least one different power policy without actually running the at least one different power policy on the computing device.