A biotechnology firm transports temperature-sensitive gene-therapy vials worldwide using reusable data loggers that record the internal container temperature every minute. Analysis of previous shipments shows that, after roughly seven days of continuous operation, many loggers read about 1.5 °C higher than a certified reference probe traveling in the same container. Engineers traced the deviation to sensor sensitivity drift. Management wants the next generation of the data-collection process to address the drift as early in the data life-cycle as possible-during the "obtain data" phase-so that downstream analytics receive temperature values that are already trustworthy. Which control should be added directly to the sensor acquisition workflow to achieve this goal?
Increase the sampling frequency to 30-second intervals so the larger data set can be averaged to smooth out drift effects.
Compress raw measurements on-device with LZ4 before transmission to reduce network latency and packet loss.
Switch the logger's output from CSV to a binary Parquet format to avoid rounding error during serialization.
Integrate an on-board in-situ calibration routine that periodically compares each logger's thermistor to a traceable reference sensor inside the shipment and automatically adjusts future readings.
Sensitivity or calibration drift is a systematic error that accumulates over time, causing the sensor to report a biased value. Because the error is systematic, taking more samples (higher frequency) or changing the file format or compression does not remove the bias; it merely stores biased data faster or in a different way. The most effective way to prevent the bias from propagating down-stream is to calibrate the sensor periodically against a traceable reference while the device is still in operation (in-situ calibration). This real-time correction updates the measurement baseline before the data leave the logger, ensuring that subsequent temperature values reflect true conditions. The other options either address data density (higher sampling), file serialization (Parquet), or transmission efficiency (LZ4 compression) but do not eliminate the underlying drift.
Ask Bash
Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.
What is in-situ calibration, and why is it important?
Open an interactive chat with Bash
Why doesn't increasing sampling frequency fix sensitivity drift?
Open an interactive chat with Bash
How do Parquet format and LZ4 compression relate to data accuracy?