Edge Device Feature Extraction for Privacy-Preserving Sensor Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Security systems generate large volumes of sensitive sensor data that require extensive computing resources for processing, and transmitting this data to remote systems raises privacy concerns.
Innovation Solution
Distribute processing of sensor data among edge devices within a local network, generating feature vectors locally to reduce data transmission and preserve privacy, with management computing systems coordinating resource allocation and privacy policies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If sensor data is transmitted to remote systems for processing, then comprehensive analysis and processing power are improved, but user privacy and data security deteriorate
Solution Approach 1:
The patent segments the processing task into two parts: feature extraction is performed locally on the edge device, while only the extracted feature vectors (not raw sensor data) are transmitted to remote systems for further processing. This segmentation maintains processing power while protecting privacy by keeping sensitive raw data local.
Solution Approach 2:
The patent extracts only the essential feature vectors from raw sensor data locally, removing unnecessary sensitive information before transmission. This extraction principle allows remote systems to perform comprehensive analysis on condensed features without accessing the original sensitive sensor data.
2Measurement precision
If all sensor data is processed centrally, then processing accuracy is improved, but network bandwidth consumption and data transmission requirements worsen
Solution Approach 1:
The patent extracts feature vectors that capture the essential information from raw sensor data, transmitting only these compact representations to remote systems. This extraction maintains processing accuracy while dramatically reducing network bandwidth consumption compared to transmitting complete sensor datasets.
Solution Approach 2:
The patent transforms raw sensor data into feature vectors, changing the data representation from high-dimensional raw measurements to condensed feature parameters. This parameter transformation preserves the essential information needed for accurate processing while reducing data volume for transmission.
3Object-affected harmful factors
If edge devices process sensor data locally, then privacy preservation is improved, but processing capability and computational resources worsen
Solution Approach 1:
The patent segments the computational workload by performing feature extraction locally on edge devices (preserving privacy) while offloading more computationally intensive analysis to remote systems with greater processing capability. This segmentation allows edge devices to maintain privacy protection without being overwhelmed by full processing requirements.
4Quantity of substance
If feature vectors are generated and transmitted instead of raw sensor data, then data transmission volume is reduced, but processing complexity at edge devices increases
Solution Approach 1:
The patent implements feature extraction functionality on edge devices to extract essential feature vectors from raw sensor data. While this adds some processing complexity at the edge, it dramatically reduces data transmission volume by sending only condensed features rather than complete raw datasets, representing a favorable trade-off.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed herein are computing system, apparatus, article of manufacture, method and/or computer program product embodiments, and/or combinations and sub-combinations thereof, for distributing processing of sensor data. For example, a computing system may be configured to determine processing capabilities of each of one or more edge devices of a network. Additionally, the computing system may determine available processing resources of each of the one or more edge devices. Moreover, the computing system may select, from the one or more edge devices, a target device, based on the processing capabilities of each of the one or more edge devices and the available processing resources of each of the one or more edge devices. Further, the computing system may communicate with the target device to cause the target device to generate one or more feature vectors based on sensor data generated by the target device.