Privacy-Aware Sensor Data Processing for Office Environment Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in obtaining accurate environmental information while preserving employee privacy and managing personally identifiable information (PII) in office settings, limiting the capabilities and flexibility of sensor data collection.
Innovation Solution
A method and system using machine learning models to process sensor data, such as camera images, to convert them into textual descriptions, removing PII dynamically based on context and user preferences, allowing for secure, efficient data storage and access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data (e.g., camera images) is collected to obtain accurate environmental information, then measurement precision is improved, but privacy risks and PII exposure increase
Solution Approach 1:
The patent extracts PII from sensor data by training the machine learning model to identify and remove personally identifiable information while retaining environmental context. This allows the system to extract only the necessary environmental information (room occupancy, desk usage) while leaving behind harmful PII elements.
Solution Approach 2:
The patent changes the parameter representation of sensor data by transforming images into textual descriptions through a machine learning model. This parameter transformation allows environmental information to be captured in a format that inherently excludes PII, resolving the contradiction between measurement precision and privacy protection.
2Loss of information
If raw sensor data is stored and processed, then information completeness is improved, but bandwidth and processing requirements increase
Solution Approach 1:
The patent creates a textual copy or representation of the visual sensor data through machine learning processing. Instead of storing and transmitting raw images, the system generates textual descriptions that capture environmental information while occupying significantly less storage and bandwidth resources.
Solution Approach 2:
The patent transforms data from one parameter format (visual images) to another (textual descriptions), reducing the information density required to represent environmental state. This parameter change maintains essential environmental information while dramatically reducing bandwidth and processing requirements.
3Object-affected harmful factors
If PII removal is applied to sensor data, then privacy protection is improved, but information accessibility may be reduced
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between sensor data collection and data storage/processing. This intermediary transforms the data into a format that inherently protects privacy while maintaining environmental information accessibility, allowing facility management to query and use the data without exposing PII.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
A method performed by an electronic device for obtaining environment information is disclosed. The method comprises obtaining sensor data from a sensor device, where the sensor data is indicative of an environment. The method comprises processing the sensor data by applying a machine learning model to the sensor data. The processing of the sensor data comprises providing a model output based on a set of parameters. Optionally, the set of parameters is configured to at least partially remove personally identifiable information. The method comprises outputting the model output. The model output is associated with time information and/or location information.