Introduction

The landscape of IoT is expanding. Every day new household and industrial devices have SMART capabilities by providing APIs to both remotely read their status and capabilities, and control their behavior. Every device has an intrinsic profile of its energy potential: what it consumes, stores and produces. Some devices only consume and store energy (like a heat-pump with large warm water reservoir), other devices can consume, store and even return energy to the grid (like some EVs). Most vendor that offer SMART remote control have their own communication protocols which puts a large effort of those who like to integrate them every time for each new use. Projects like InterConnect and Hedge-IoT have the goal of making integration easier by proposing a shared communication layer where both the syntax and semantics are agreed upon. The only effort then for each vendor is to make a translation of its own language to an open semantic standard like SAREF. After this each integrator can easily communicate with all these devices.

The ETSI SAREF stack covers a large domain of things related to the energy and IoT domain, for example ways to describe sensor measurements, locations in buildings, energy consumption and production profiles, functional triggers (e.g. “On/Off”), and much more. To demonstrate a real-life scenario where a SMART office with numerous sensors from various vendors is translated to SAREF is the OfficeGraph. Where this result only covers sensors, the results are a success and will be extended within the Dutch Pilot of HEDGE-IoT project, where we will also translate the data and functionality of various devices like Ev chargers, batteries, solar panels, heat-pumps and more.

Main Body

The figure below illustrates the number of translations are needed in two situations where all devices need to communicate with each other: 1) (n*n-1) /2 without a shared language and 2) n with a shared language (where n are the number of devices).
[figure 1]
As mentioned, the OfficeGraph is using the SAREF standards to describe the various measurements, locations and devices. SAREF in itself is an ontology, allowing semantically rich descriptions using Linked Data as the underlying paradigm.

Linked Data is a method of structuring data so that it can be interlinked and easily shared across different systems. It uses standard web technologies like HTTP, RDF, and URIs to connect data points, making it possible for machines to understand and query the relationships between them,  [1,2] . This approach transforms the web into a global database, enhancing the accessibility and usability of data.

OfficeGraph is a large, real world knowledge graph containing measurements taken by 444 IoT devices, over 11 months, in a seven story office building in Eindhoven, The Netherlands. The devices are made up of 17 different sensor models, which make measurements of many different properties.

The main structure we used from SAREF can be seen in figure 2:

The devices in room enrichment adds more information about which devices are located in which rooms, and on which floor those rooms are.

The Wikidata days enrichment provides a link to Wikidata, by matching the dates of the measurements to those dates’ entities in Wikidata.

The graph learning enrichment provides additional properties that have beneficial effects on the learning process when using graph embedding models.

The enrichments are located in separate files, with the graph learning enrichment folder only containing the enrichments for the devices on the 7th floor, which were used in a machine learning experiment (the code of this experiment is available on GitHub [3]).

The resource paper describing this dataset is published at the ESWC 2024 conference[4].

Now that we have this realistic dataset in SAREF format, we can demonstrate its use in various ways.

For example, by using Jupyter Notebooks, we can visualize aggregated data based on specific questions like “what is the relation to thermostat measurements and if windows are open or not?”.

This demonstration can be found here [5].

Secondly, we developed a knowledge-based home automation system in which scenarios are the result of logical inferences over the IoT sensors data combined with formalized knowledge [6].

Thirdly, in the context of Hedge-IoT we will use the OfficeGraph as a dataset to develop and test the AI graph learning algorithm to detect outliers which can be used for predictive maintenance of the devices. This work is based on research by Wilcke et al. [7] and will be applied in the context of the Dutch Pilot within the HEDGE-IoT project.

Conclusion

“A little semantics goes a long way” (James Hendler 2001). Semantic Web technology and Linked Data is nowadays a very mature field and demonstrated its use in various domains. In our context of IoT and the goals of the energy transition, it is paramount that the integration of, and communication between, the various devices and systems is streamlined. Linked Data is specially developed to both have a shared syntax and semantics combined with automated inference possibilities. The OfficeGraph is a real life Linked Data set which can be used as a benchmark dataset to test and demonstrate the possibilities in the context of Hedge-IoT. Within the Dutch Pilot in HEDGE-IoT we plan to create the CampusGraph, which will contain even more elaborate data in SAREF and create new demonstrations of its use.

References

[1]  https://en.wikipedia.org/wiki/Linked_data

[2]  https://data.europa.eu/en/publications/datastories/linking-data-what-does-it-mean

[3] https://github.com/RoderickvanderWeerdt/OfficeGraph

[4] https://dl.acm.org/doi/10.1007/978-3-031-60635-9_6

[5] https://github.com/RoderickvanderWeerdt/OfficeGraph/blob/main/data%20analytics%20use%20cases/Competency%20Question%201%20-%20thermostat%20and%20windows.ipynb

[6] Reda R, Carbonaro A, de Boer V, Siebes R, van der Weerdt R, Nouwt B, Daniele L. Supporting Smart Home Scenarios Using OWL and SWRL Rules. Sensors. 2022; 22(11):4131. https://doi.org/10.3390/s22114131 

[7] User-centric pattern mining on knowledge graphs: An archaeological case study WX Wilcke, V de Boer, MTM de Kleijn, FAH van Harmelen, HJ Scholten

Authors:

Roderick van der Weerdt, Victor de Boer, Ronald Siebes, Xander Wilcke (Vrije Universiteit Amsterdam)