Sensor Reading Reliability Using Learned Error Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor systems lack effective methods to accurately determine the reliability of sensor readings, particularly in detecting malfunctioning or spoofed sensors, which can lead to inaccurate process control in physical processes like manufacturing and baking, where sensor failures or tampering can result in suboptimal outcomes.
Innovation Solution
A computing device-based system that learns a sensor signature from historical data to determine a sensor's error rate and rate of change, allowing for the comparison of current readings to assess their reliability, thereby identifying potential sensor malfunctions or tampering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical restrictions are used to protect sensor hardware access, then security is improved, but the system cannot detect when sensor readings are spoofed
Solution Approach 1:
The system performs preliminary actions by learning the sensor's normal behavior patterns (error rate and rate of change) during a training phase before actual monitoring begins. This pre-established baseline enables the system to detect anomalies without requiring complex real-time analysis mechanisms.
Solution Approach 2:
The patent introduces an intermediary computational layer that analyzes sensor readings against learned patterns. This intermediary system (the reliability determination mechanism) mediates between the physical sensor and the control system, detecting spoofed readings without requiring direct modification of the sensor hardware or complex physical security measures.
2Reliability
If sensor spoofing detection is added to simple sensors, then detection capability is improved, but processing power and real estate requirements increase
Solution Approach 1:
The patent extracts the complex detection logic from the sensor itself and places it in an external computing device. The sensor remains simple, only providing raw readings, while the spoofing detection functionality is taken out and implemented separately through software-based analysis of the sensor's behavioral patterns.
Solution Approach 2:
The system creates a virtual model (copy) of the sensor's normal behavior through learned patterns of error rate and rate of change. This behavioral copy enables detection of anomalies without requiring the actual sensor to have complex detection capabilities, as the analysis is performed on copies of the sensor data rather than the sensor hardware itself.
3Productivity
If traditional sensor monitoring is used, then system operation is maintained, but inaccurate readings from malfunctioning sensors go undetected
Solution Approach 1:
The system implements feedback by continuously comparing current sensor readings against the learned sensor signature patterns. When deviations exceed expected error rates or rate of change thresholds, the system generates feedback signals indicating potential spoofing or malfunction, enabling real-time correction or investigation of inaccurate readings.
Data Source
AI summary
Examples include receiving a plurality of values of a parameter that is measured by a sensor, determining a sensor rate of change based on the plurality of values, determining a sensor error rate based on the plurality of values, receiving a reading of the parameter, and determining a reliability of the reading based on the sensor rate of change and the sensor error rate.