Accurate Sample Time Reconstruction of Inertial FIFO Data

Area of Science:

• Cyber-physical systems
• Sensor data processing
• Micro-electromechanical systems (MEMS)

Background:

• Accurate sensor data is vital for sensor data fusion and feature extraction in modern cyber-physical systems.
• Simultaneously connected sensor devices pose challenges for energy-efficient data acquisition and processing.
• First-in-first-out (FIFO) interfaces in sensors help manage multiple data samples but amplify clock drift issues.

Purpose of the Study:

• To develop an accurate sample time reconstruction method for sensor data.
• To address the challenges of clock drift and timing offset errors in systems with multiple sensors.
• To achieve robust and energy-saving data acquisition and processing.

Main Methods:

• Utilizing an internal sensor timer available in MEMS technology sensors.
• Implementing a forward-only processing approach for timestamp calculation using the sensor FIFO interface.
• Developing an algorithm for accurate sample time reconstruction independent of clock drift.

Main Results:

• Achieved a standard deviation of reconstructed sampling periods below 40 μs.
• Demonstrated run-time savings of up to 42% compared to single sample acquisition.
• Provided a robust and energy-efficient solution for sensor data timestamping.

Conclusions:

• The proposed approach enables accurate sample time reconstruction, mitigating clock drift effects.
• The method offers significant energy and run-time savings in sensor data processing.
• This technique enhances the quality of sensor data fusion and feature extraction in cyber-physical systems.