Ambient sensor prediction pipeline
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional ambient sensing technologies face privacy challenges due to the use of personally identifiable information (PII) and facial features, which expose sensitive information, and are inefficient with continuous sensor data transmission leading to computing resource waste.
Innovation Solution
A device configuration with local filtering and network-connected prediction systems processes sensor data locally based on relevance, using tracking target signatures from point cloud features, reducing data transmission and enhancing privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor data is continuously transmitted to remote system for processing, then monitoring coverage is maintained, but computing resource waste and network inefficiency occur
Solution Approach 1:
The system performs preliminary filtering of sensor data at the sensing device before transmission. A local filter mechanism evaluates sensor data against predefined criteria and only transmits data that meets relevance thresholds, preventing unnecessary network traffic and remote processing of irrelevant data.
Solution Approach 2:
The monitoring system is segmented into two functional parts: local filtering at the sensing device and remote processing only for relevant data. This segmentation allows the system to maintain monitoring coverage while reducing the computational burden on remote systems by handling preliminary data evaluation locally.
2Measurement precision
If PII and facial features are used to identify occupants, then identification accuracy is improved, but privacy security deteriorates
Solution Approach 1:
The system extracts only the necessary identifying features (such as gait patterns, spatial-temporal characteristics) from sensor data while deliberately excluding personally identifiable information and facial features. This extraction approach maintains identification functionality while removing privacy-sensitive data elements.
Solution Approach 2:
The system uses temporary, non-PII identifiers that are generated and discarded rather than storing persistent personal information. These short-living identifiers enable tracking and identification without creating long-term privacy vulnerabilities or exposing sensitive personal data.
3Loss of information
If all sensor data is transmitted to remote system, then data completeness is maintained, but network traffic and processing load increase
Solution Approach 1:
The system applies partial action by transmitting only a subset of sensor data that meets relevance criteria rather than all collected data. The local filter mechanism selectively transmits data based on predefined thresholds and conditions, achieving sufficient monitoring coverage without the overhead of complete data transmission.
Solution Approach 2:
Data filtering and relevance evaluation are performed preliminarily at the sensing device before transmission. This preliminary action ensures that only data meeting specific criteria is transmitted, maintaining necessary data completeness for monitoring purposes while significantly reducing network traffic and remote processing requirements.
Data Source
AI summary
Various embodiments of the present disclosure provide an ambient sensor prediction technique that improves the functionality of a computer in various aspects. The technique comprises receiving an excursion message that comprises sensor-based feature values, identifying an entity signature for the excursion message based on a first subset of the plurality of sensor-based feature values and one or more contextual attributes based on a second subset of the plurality of sensor-based feature values, identifying a target log file corresponding to the excursion message based on a comparison between the entity signature and a tracking target signature corresponding to the target log file, and storing the one or more contextual attributes as one or more of a plurality of historical contextual attributes of the target log file.