The Institute for Data Engineering is currently conducting research in the following areas, as well as in related areas. Information on ongoing and past projects can be found at the project website, while our scientific papers are listed on our publication website

Blockchain Technologies for the Internet of Things

In the Christian Doppler Laboratory Blockchain Technologies for the Internet of Things (CDL-BOT), researchers work on different research questions which aim to make Distributed Ledger Technologies (DLTs) suitable for applications in the Internet of Things (IoT). 

The research problems taken into account range from blockchain interoperability to lightweight DLTs and support for software developers. CDL-BOT aims at making DLTs a commodity in the IoT - for applications ranging from trustful data exchange and data storage to payment mechanisms.

CDL-BOT is located at TU Hamburg and TU Wien. The laboratory is financed by Austria's Federal Ministry for Digital and Economic Affairs and its industrial partners Pantos (Austria) and the IOTA Foundation (Germany).

Federated Learning

Today's machine learning approaches typically rely on large amounts of data being sent to a centralized component, which is very often located in the cloud. Apart from the fact that organizations and private persons are often not very keen on sharing raw data in an intransparent way, such centralized approaches to machine learning may also lead to a too large communication overhead.

In our research, we therefore investigate novel approaches to artificial intelligence at the edge, applying federated learning for this. In federated learning, machine learning is done in a distributed way by independent nodes. Only after a local model has been generated, the model is shared with other nodes, thus decreasing the amount of data to be exchanged by several magnitudes. Also, raw data does not need to be shared at all, thus increasing data privacy.

In our research, we investigate novel ways to minimize the communication overhead in federated learning environments while still achieving excellent learning results. Also, the identification of meaningful learning "cohorts", i.e., groups of nodes which collaborate in federated learning, is one of our research directions. While the focus of this research is on industrial settings, the outcomes can be applied to arbitrary smart systems.