Sensor Technology

Combined with digital 3D city models, the sensors provide dynamic input to create accurate and up-to-date virtual representations of cities. This integration offers a comprehensive understanding of the urban environments, enabling city planners, policymakers, and citizens to monitor conditions, simulate scenarios, and make informed decisions that enhance sustainability, resilience, and the overall quality of urban life. Integrated analysis and visualisation of 3D city models and sensor data enables city stakeholders to intuitively engage with both spatial and temporal dimensions simultaneously, allowing them to analyse citywide trends.

Urban sensor data originates from various sources, each with distinct data models and exchange formats, which complicates the utilisation of the data in Urban Digital Twins applications. For standardised management of heterogeneous sensor data, the OGC SensorThings API offers a standardised framework for integrating and managing IoT devices and data (Liang et al., 2021). It enables the interoperability of interconnected IoT devices by utilising a standard data model and a RESTful API for accessing and querying sensor data. The standard is implemented in software offerings, the most popular being the Fraunhofer Open Source SensorThings API (FROST) Server by Fraunhofer IOSB. An IoT stack developed by the Chair of Geoinformatics at the Technical University of Munich builds on open-source standards and tools, including the FROST Server for integrating, managing, and visualising heterogeneous IoT sensor data (Willenborg et al., 2024). The stack offers flexible deployment options, enabling a choice between Docker or a Kubernetes cluster, tailored to the specific needs of the sensor data infrastructure.

Data integration remains one of the significant challenges in creating Urban Digital Twins. The CityGML 3.0 Dynamizers module offers a standardised framework for linking time-dependent properties of semantic 3D city models to temporal data sources (Kolbe et al., 2021). These temporal data sources include IoT platforms, databases, or external tabular files. The CityGML Dynamizers module is fully supported in the schema of the open-source software 3DCityDB v5, enabling the storage and exchange of rich, semantic 3D city models, along with their corresponding temporal data sources.

Web visualisation of integrated 3D city models and sensor data requires transferring the data structures defined by CityGML Dynamizers to web-compatible streaming and rendering formats such as OGC 3D Tiles and OGC I3S[AD1] . Within the CUT project, we are developing a framework to bring CityGML Dynamizers to the web, enabling web applications to discover and interpret the underlying data structures for 4D web visualisation in Urban Digital Twins. The linked web application showcases the visualisation of 3D city models and sensor data integrated using CityGML Dynamizers (4D Web Visualisation Demo).

Urban metadata catalogues provide structured systems for describing, indexing, and managing metadata associated with various urban datasets. These catalogues help users discover, access, and reuse urban datasets, such as data from IoT sensors, across various applications, including urban digital twins. Currently, registering sensor data in metadata catalogues is time-consuming and relies on human intervention, making it prone to errors and resulting in outdated entries as services evolve. Urban IoT services that incorporate hundreds or thousands of individual sensors necessitate substantial grouping and high-quality metadata to ensure discoverability for diverse user groups, including municipal agencies, city planners, and transportation departments. To tackle these challenges, within the CUT project, we propose a framework for the continuous automated registration of IoT sensors in metadata catalogues to improve their discoverability. This framework comprises three main steps: harvesting sensor data from IoT services, categorising the sensors, and creating entries based on the derived categories. The framework leverages large language models (LLMs) to group sensors and generate enriched metadata entries, thereby enhancing the discovery of sensor data (Limnardy, 2025).

Kolbe, T. H., Kutzner, T., Smyth, C. S., Nagel, C., Roensdorf, C., & Heazel, C. (2021). OGC City Geography Markup Language (CityGML) Part 1: Conceptual Model Standard (Version 3.0). Open Geospatial Consortium. https://docs.ogc.org/is/20-010/20-010.html

Liang, S., Khalafbeigi, T., & van der Schaaf, H. (2021). OGC SensorThings API Part 1: Sensing Version 1.1 (Version 1.1). Open Geospatial Consortium. https://docs.ogc.org/is/18-088/18-088.html

Limnardy, J. (2025). Automated classification of IoT sensors and registration in urban data catalogs [Masters, Technical University of Munich]. https://mediatum.ub.tum.de/node?id=1796206&change_language=en

Willenborg, B., Kolbe, T. H., & Schwab, B. (2024). tum-gis/tum-gis-iot-stack-k8s: tum-gis-iot-stack-k8s-0.11.0-beta3 [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.13950355