News

Reliable and accurate methods for data quality in sensor networks

Smart city

Lowering measurement uncertainty and improving SI-traceability

Sensor networks can be comprised of dozens, hundreds, or thousands of sensors, often measuring different parameters under different environmental conditions. They can be used to ensure conformity with European Directives on emissions, are able to control processes, and monitor quality control in advanced manufacturing. Sensor networks are also used to measure pollution in cities and are integral to the Internet of Things.

However, developments in sensor types and the introduction of artificial intelligence software systems have meant that these networks are struggling with providing consistent data quality. In addition, many networks have unknown measurement uncertainties and lack traceability to the SI system of units.

To overcome these issues, Metrology Partnership project Fundamental principles of sensor network metrology (22DIT02, FunSNM) is working to address the metrological aspects of sensor networks, covering uncertainty propagation, data quality metrics and SI-traceability. In addition, this project will cover the assessment, infrastructure, and risk analysis of distributed sensor networks alongside software frameworks by developing automated applications. The applicability of the methods, tools, and concepts will be demonstrated in typical real-world sensor networks.

Guidance document on data quality metrics in sensor networks

A new Good practice guide for developing agent-based sensor networks has been published. The guide includes practical approaches for implementing data quality metrics and ensuring SI-traceability in diverse real-world scenarios. Developed collaboratively by project partners, the document is expected to support end users in improving data reliability and optimising calibration strategies across sensor networks.

Other project developments:

Other project developments so far include:

Open access paper Measurement uncertainty evaluation for sensor metrology detailing new methods for evaluating uncertainty in real-world sensor networks. These techniques, developed during the first year of the project, enable more accurate, SI-traceable measurements even in complex and dynamic environments.
Investigation into the correlation of sensor errors. Using data from over 600 sensor units in Paris and Helsinki, researchers developed Gaussian process models to assess how error correlations affect measurement uncertainty. These insights will help improve the accuracy of air quality estimates.
Successfully applied Laplace domain tools (converts a function from the time domain to the frequency domain) for the co-calibration of air quality sensors in a dynamic urban environment. Using a network of sensors deployed across Helsinki, researchers calibrated temperature, humidity, pressure, and particulate matter sensors in situ. This innovative approach enables real-time adjustment of low-cost sensor networks, enhancing both accuracy and reliability, and lays the foundation for future standards in environmental monitoring.
Development of methods to assess and improve data quality in sensor networks monitoring district heating systems. By simulating meter error estimation and analysing the impact of sensor placement, the team demonstrated how additional measurement points can significantly enhance system accuracy. These insights are expected to support utilities and city planners in optimising heating infrastructure for energy efficiency and reliability.
Development of drift mitigation methodologies for thermocouples used in extreme industrial environments. In collaboration with industry, project partners exposed multi-wire thermocouples to temperatures above 1500 °C, performing periodic in-situ recalibrations. This work has led to the creation of multiple drift correction techniques based on physical, data-driven, and hybrid models, significantly improving the reliability of temperature measurements. These advancements are expected to benefit high-value manufacturing sectors, such as aerospace and metal heat treatment, where accurate thermal control is critical.

At the end of the project, information on how to lower the measurement uncertainty and establish SI-traceability for static and mobile sensor networks will positively impact environmental monitoring and help support European Directives.

Project coordinator Shahin Tabandeh from VTT said

‘These developments show how metrology can give sensor networks the reliability, traceability, and uncertainty awareness needed to support better decisions in real-world applications. Our aim is to ensure that future networks provide trustworthy data together with their associated measurement uncertainty, supporting both effective monitoring and more efficient industrial processes across Europe.’

This Metrology Partnership project has received funding from the European Partnership on Metrology, co-financed by the European Union Horizon Europe Research and Innovation Programme and from the Participating States.

Want to hear more about EURAMET?