A search engine for the Internet of Things

We are familiar with web crawlers in the form of the efficient search engine bots as used, for example, by Google. Their name comes from the fact they “crawl” through the entire visible web and index the contents. An international research project has now come up with something similar for the Internet of Things (IoT): the “IoTCrawler.” Siemens has developed a prototype capable of seeking out flexible electricity generators or consumers.


If a power network experiences an overload or shortfall – in other words, if too little or too much electricity is being generated or consumed at a given moment, the grid operators have to take countermeasures to prevent disruptions. “Previously, fluctuations of this nature were offset by power stations or industrial consumers,” explains Andreas Fernbach, of the IoT research group at Siemens Austria. “But in the future, decentralized elements will also need to be brought in for this purpose – that means the majority of households. Large home battery storage systems, for example, or wall boxes for charging electric vehicles, photovoltaic inverters, and heat pumps.” Every household that wants to take part can put its consumers or generators to work in the market mechanisms and help keep the power network stable. That means the controller needs to be familiarized with the participating resources in advance. Then, depending on requirements and availability, the appropriate devices will be sought from this pool and activated at the correct time.

Research project for an “IoT Google”

The international research project "IoTCrawler“ created a technical framework for efficiently searching devices in the IoT. The IoT operates using a range of applications, and in the future will constitute a universal platform for scalable discovery applications subject to the observance of security and data protection regulations. “As a precondition for the crawler framework, we first had to define and integrate the relevant IoT assets,” relates Josiane Xavier Parreira from Siemens Research in Austria. “Unlike the situation with the Internet, search objects in the IoT don’t actively make themselves visible to the search engine; the various assets must first be described so that the applications building on the crawler understand which possible selections exist, and with which properties, before indexing can take place.”

The IoT search machine forms a base the various applications can build on. “Depending on the case, the IoT assets used in the search process will also differ. For our purposes it was flexibility in electricity grids. For other applications it may involve air quality sensors, machine run times, or available parking spaces,” says Xavier Parreira, who headed the project for Siemens.

Ontologien and Knowledge Graphs

“The IoTCrawler system uses ontologies and knowledge graphs in order to recognize and understand the asset landscape,” continues Xavier Parreira. “These are standard methods for describing knowledge and relationships. The ontologies describe the relevant assets and thus ensure that all participants have the same understanding of them, for example whether a battery is a component that has capacity and a maximum permissible charging power. Knowledge graphs are used to represent and structure the connections at a content level. When represented in graphic form, this can look like a map of the underground railway system or mindmaps. Semantic enrichment of the data is also important from a methodical perspective, in other words, the allocation of meaningful definitions to sensor values such as physical size or reliability, for example.” The project also drew on methods of Machine Learning, which were used to interpret the data flow generated by the sensors.

How things get into the Internet

There is still the question of how things get into the Internet in the first place in order to form the Internet of Things. Fernbach answers this question based on the Siemens use case: “A practical approach is to make use of the manufacturer’s existing infrastructure. For example, PV inverter or battery storage system manufacturers have already established a monitoring infrastructure; in other words, their devices are already in the Internet and are supplying data to manufacturers. In this way it’s possible to access a specific pool of assets.”


Available as Open Source

The IoTCrawler components were made available to the Open Source community after the project ended. In some cases, use case concepts from partner entities are being pursued further. The prototype created by Siemens as part of the project is now available for further in-house projects and research projects. Siemens is testing its use case scenario for further practical applications. “We’ve developed the search functionality to the point where it’s now available for other application domains,” Xavier Parreira concludes.

Christian Lettner (hi!tech), Oktober 2021

Subscribe to our Newsletter

Stay up to date at all times: everything you need to know about electrification, automation, and digitalization.