Large-Scale Virtual Calibration of MEMS Sensors

The EU Digital Decade Policy Programme 2030 strongly depends on safe and reliable cutting-edge technologies, like Micro-Electro-Mechanical Systems (MEMS) sensors, that are widely used in large sensor networks for infrastructural, environmental, healthcare, safety, automotive, energy and industrial monitoring. Their massive production, often in the order of millions/week, requires costly and time-consuming calibration processes, resulting in a lack of metrological traceability. Indeed, current standard calibration procedures require that each sensor undergoes an in-the-lab individual calibration, which, for the new-generation MEMS sensors, is unfeasible. Consequently, there arises a critical need to establish comprehensive calibration methods on a large scale for this kind of sensors and related sensor networks, which ensure adherence to standard technical procedures and impartiality. In particular, it is fundamental that a systematic metrology framework for trustworthy and safe calibration of digital sensing technologies on a large scale is developed and implemented. A solution for large-scale virtual calibrations of MEMS sensors, which was recently proposed and is now under further development, lies in the use of virtual calibration methods employing Bayesian statistical tools. These methods are able to substitute or complement traditional in-the-lab calibrations and can drastically reduce time and cost efforts while providing traceability at acceptable uncertainty and reliability levels. The proposed Bayesian framework allows to statistically calibrate large batches of sensors using probabilistic models and prior knowledge. It involves the experimental (in-the-lab) calibration of only a small sample of sensors drawn from a large batch, then it infers the number of reliable sensors in the entire batch and assigns an appropriate uncertainty value to all the sensors. Therefore, it can be considered as a statistical calibration of the batch. The Bayesian nature of this approach allows reducing the number of experimental calibrations by incorporating the prior knowledge coming from the previous calibration of a ‘benchmark’ batch, representative of the whole production process, which is performed ‘once and for all’. Strategies are under study to balance the costs and time effort faced in the virtual calibration and the desired levels of batch reliability and uncertainty. This shift of paradigm from individual device calibration to statistical sampling is also highlighted in the BIPM CCAUV's 2021-2031 strategy and EMN Mathmet's 2023-2033 Strategic Research Agenda. Such methods for large-scale virtual calibration will allow accredited calibration laboratories to apply standard procedures and attribute sensitivity and uncertainty to millions of MEMS sensors, ensuring their reliability. They will serve to industries, such as semiconductor and MEMS manufacturers, sensor network suppliers and end-users, standards developing organisations and NMIs and DIs to ensure traceability, accuracy and trustworthiness of their sensor measurements. The enhanced efficiency of the proposed calibration will result in cost and time savings in product development for European manufacturers. Improved traceability will strengthen competitiveness in the MEMS sensor industry, which is expected to increase from € 16.3 billion in 2021 to € 35.9 billion by 2030 due to demand in areas such as automotive, healthcare, and industrial automation. Traceable MEMS sensor networks will improve public health, safety, and urban development.