Sensor Time Stamp Synchronization Using Regression Drift Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Sensor time stamps drift from application time stamps due to different time domains, leading to inaccurate integration of sensor data samples and potential failure of sophisticated functions, especially in systems with operating systems where interrupt service routines assumptions are invalid.
Innovation Solution
A regression model is generated using fixed point arithmetic to estimate application time stamps from sensor time stamps, periodically updating parameters to track drift between time domains, ensuring accurate integration of sensor data samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If interrupt service routines are used for time stamp mapping, then the solution works in low jitter systems, but it fails to scale to operating system kernels where execution timing is unpredictable
Solution Approach 1:
The patent replaces the mechanical/timing-based interrupt service routine approach with a software-based regression model that uses mathematical calculations to estimate application time stamps. This substitution eliminates dependency on predictable execution timing and allows the solution to work across different system architectures including operating system kernels.
Solution Approach 2:
The patent changes the approach from direct time stamp mapping (assuming equal execution times) to parametric estimation using regression models with multiple parameters (slope, intercept, offset). These parameters are calibrated to account for systematic timing differences, enabling accurate time stamp conversion without relying on predictable execution timing.
2Measurement precision
If sensor time stamps are directly used, then the sensor data is timestamped in real-time, but the time stamps drift from application time stamps over time due to different clock sources
Solution Approach 1:
The patent implements feedback by continuously monitoring the difference between sensor time stamps and application time stamps, then using this information to update regression model parameters. This feedback loop compensates for clock drift and maintains accurate time stamp conversion over extended periods.
Solution Approach 2:
The patent performs preliminary calibration by collecting time stamp data pairs and fitting regression models before actual time stamp conversion begins. This preliminary action establishes the relationship between sensor and application time domains, accounting for systematic offsets and drift rates before they affect measurement accuracy.
3Adaptability or versatility
If multiple sensors with different time domains are integrated, then more sophisticated functions can be provided, but time stamp accuracy becomes increasingly problematic
Solution Approach 1:
The patent creates a universal time stamp conversion framework using regression models that can handle multiple sensors with different time domains through a single unified approach. The same regression model infrastructure adapts to each sensor type, providing consistent accurate time stamp conversion across diverse sensor sources without requiring sensor-specific timing mechanisms.
Data Source
AI summary
A device includes processing circuitry and sensor communication interface circuitry. The processing circuitry, in operation, generates one or more parameters of a regression model based on a plurality of data sets. Each data set includes an application time stamp and a sensor time stamp. Based on the one or more parameters, the processing circuitry estimates application time stamps associated with received sensor data samples. The received sensor data samples have associated sensor time stamps. The sensor communication interface circuitry is coupled to the processing circuitry. The sensor communication interface circuitry, in operation, communicatively couples the processing circuitry to a sensor. Data samples of the sensor are received by the processor via the sensor communication interface circuitry.