The conformity of modern measuring instruments regulated with legal requirements is usually demonstrated by verifying software version numbers and cryptographic hashes over executable binaries. Regulated measuring instruments in Legal Metrology could also pass this conformity test, if the functional behavior of devices in the field and certified prototype is identical, thereby giving rise to the need for more efficient monitoring and identification methods in many legally regulated industries, especially when the functionalities of devices remain hidden from the view of authorities, i.e., in black box scenarios. In this paper, we extend a risk-based software monitoring algorithm by developing and integrating a model learning algorithm that works in quasi-black-box settings. We also analyze the performance of the algorithm and compare it with state-of-the-art approaches. If successfully implemented, manufacturers of measuring instruments could introduce software updates and monitor potential abnormalities, without resubmitting the software for new conformity assessment or certification.
Themenbereich der Metrologie
Mathematik und metrologische Informationstechnik
Innovationscluster
Digitalisierung
; Systemische Metrologie
Geschäftsfelder
Metrologie für die Gesellschaft
; Metrologie für die Wirtschaft
Zitierung
Ho, L. C. X., Esche, M., Nischwitz, M., & Glesner, S. (2025). Black-box conformity tests on regulated measuring instruments: A machine learning approach. 2025 IEEE International Instrumentation and Measurement Technology Conference (I2MTC), Chemnitz, 19-22, Mai, 2025, Deutschland. https://doi.org/10.1109/I2MTC62753.2025.11079177
