Sensor Optimization for Model Accuracy and Battery Life
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing use of smart applications and machine learning models on devices like smartphones leads to rapid battery drain and data storage depletion due to frequent data acquisition and transmission, which affects device performance and user experience.
Innovation Solution
A computer-implemented method and device that optimize sensor settings to balance model accuracy with battery consumption and data storage space, involving the determination of optimal sampling configurations for sensors, such as GPS, cameras, and microphones, to minimize power usage while maintaining desired model performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data acquisition frequency is increased to improve model accuracy, then model accuracy is improved, but battery consumption increases
Solution Approach 1:
The patent implements dynamic sensor configuration that adapts sampling rates based on current operational context and model performance requirements. The system continuously adjusts data acquisition frequency, enabling high accuracy when needed while reducing battery consumption during periods when lower accuracy suffices, thus resolving the contradiction between model accuracy and battery consumption
Solution Approach 2:
The system changes operational parameters of sensors (sampling rate, resolution, activation state) based on model accuracy requirements and battery status. By dynamically adjusting these parameters, the system maintains adequate model accuracy while optimizing battery consumption, directly addressing the technical contradiction
2Measurement precision
If data acquisition frequency is increased to improve model accuracy, then model accuracy is improved, but data storage space is depleted
Solution Approach 1:
The patent extracts and transmits only the most relevant features and processed data to remote servers, rather than storing all raw sensor data locally. This selective extraction approach maintains model accuracy by preserving critical information while significantly reducing local data storage requirements, thus resolving the contradiction between model accuracy and data storage space
3Measurement precision
If sensor activity is increased to acquire large amounts of data, then model accuracy is improved, but device performance degrades
Solution Approach 1:
The patent segments data processing into multiple stages: initial filtering and feature extraction on the device, selective transmission to remote servers, and further processing in the cloud. This segmentation reduces the computational burden on the device, maintaining model accuracy while preserving device performance for other applications
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a computer-implemented method for determining sensor settings of a device to optimize a trade-off between model accuracy, battery consumption and data storage amounts. The method comprises iteratively assessing different sensor settings for the model accuracy achieved based on these settings taking into account the associated cost in terms of battery consumption and data storage space. Once a model has been identified that offers a desirable trade-off, the device is configures to operate sensors with the setting(s) associated with the identified model.