protect and connect valuable data
Why Federated Computing?
Data is the “new oil”. It fuels the transformation of entire industries. It powers the digital economy. Enterprises and consumers begin to understand the value of their data assets. Society and science rely on data to make informed decisions. In all sectors, information becomes more valuable and more reliable as data is pooled from inaccessible silos into larger and more accessible data lakes.
But there is a lot of friction. Data sharing is often not an option. Companies need to protect their proprietary data. Consumers are concerned about privacy. Data protection and data security are ubiquitous requirements. Informational self-determination is seen as a human right. Data lakes and large scale data bases become prime hacking targets. Trustees, contracts and licenses are costly. There seems to be a fundamental tradeoff between value and efficiency vs. privacy and security.
There are modern solutions. Using encrypted peer-to-peer networks and cloud federation, data owners may retain complete control over when, how, and by whom their data is used. Even in anonymous settings or in the absence of trust, collaboration becomes possible again. Data owners may reap all the benefits of aggregating their data without actually sharing it with others.
collaboration without data sharing
How do Federated Computing technologies work?
Secure Multiparty Computing (SMPC, MPC): The gold standard with mathematical security proofs. Participants exchange encrypted messages in peer-to-peer networks. The network protocol ensures that only the intended result of the joint computation becomes public. All input data remains strictly on premises with their original owners. Protocols may be hardened to protect against malicious parties. Military grade security is achievable against attackers with resources of nation states. Applications in defense, critical infrastructure, healthcare, finance and e-government.
SMPC Success Story:
We are pioneers of federated computing in the European health data space. Our team completed the first SMPC computation with live patient data in Germany in 2019 and across European borders in 2024.
Federated Learning (FL): AI models are trained on distributed data. Instead of pooling the data and central training (“data to code”), the AI model is sent around and trained locally (“code to data”). Over time, the AI model learns the features in all of the data sets without the need to share the data sets openly. This can help to train more accurate models and to reduce learning bias by including data from diverse sources. Applications in e-commerce, automotive, medical research, large language models, generative AI.
Fully Homomorphic Encryption (FHE): Data is locally encrypted and then sent to a 3rd party for processing. The 3rd party may be a cloud hyperscaler or another infrastructure provider who is not supposed to have access to the data. Applications in big data, pharma, and virtual private cloud (VPC) infrastructure.
Differential Privacy (DP): Researchers are allocated a “privacy budget” and are only allowed to make limited requests to a distributed data base. Requests are only fulfilled as no individual data items are revealed. Applications in medical research and other fields of sensitive individual data.
consulting, architecture and solutions
What do we offer?
Consulting. We have been advising the public and private sector on multiple occasions. From market scans to recommendations on technology, policy, process, and implementation, we provide superior insights to institutions and companies transitioning to federated computing. As outside experts, we strengthen the research of consulting firms or directly join their engagement teams on occasion.
Architecture. As external experts, we advise your architects and developers how to best integrate federated computing into your infrastructure whether you are using hyperscaler clouds or servers on premise. We help your team make the correct design choices and plan architecture and DevSecOps accordingly. Our recommendations are provider independent and product neutral.
Architecture Success Story:
Our free and open source architecture and reference implementation of “Federated Secure Computing” developed for LMU Munich won the Innovation Award by Stifterverband.
Solutions. We leverage industry tested 3rd party tools and enterprise software to deliver your use case the best way possible. We implement core functionality for bespoke applications, ranging from simple one-off computations between two or three partners to high-throughput real-time cross-region and multi-cloud Kafka pipelines connecting millions of IoT devices in redundant and secure ways.
fdrtd: We brand our proprietary solutions as “fdrtd” – short for “federated”. fdrtd is a registered trademark of bytes for life.