How data spaces could scale creating digital twins

Building a digital twin of a product or a factory involves coordinating many moving parts into one coherent whole. But at least most of this data is stored somewhere within the enterprise. Building scalable digital twins to represent cities, food systems, logistics and energy grids will require cooperation between many participants and even competitors. 

Engineering, industrial and government collaborators are starting to make progress on data spaces – an emerging concept for aligning the technical, business and legal aspects required to share data. At the Imec Future Summits conference, Tanguy Coenen explained how data spaces provide the foundation to scale digital twins across organizations in an exclusive interview with VentureBeat. Coenen is a digital transformation strategist at Imec, a cross-industry scientific collaboration in Europe.

He has been involved in building local digital twins of Antwerp and working with a cross-disciplinary team to extend data spaces across different industries to improve sustainability in food, energy and logistics. Early research focused on the practical aspects required to share sensitive data, negotiate legal agreements and ensure business competitiveness. Now his group is working with enterprises to scale these efforts on top of commercial tools and services. Eventually, this could help create a marketplace of interoperable digital twins and connected machine learning models.

Building local digital twins

A few years ago, Coenen helped coordinate research on a digital twin of the city of Antwerp. The goal was to help public sector decision-makers improve their understanding and make better decisions. They found that most decisions involved multiple kinds of data. For example, efforts to improve traffic will impact pollution, commerce and citizens. 

“We found that people that make decisions in cities always make cross domain decisions,” Coenen said. 

Imec collaborated with local partners to build prototypes of how this might work as a proof-of-concept to inspire industry and governments on how a local digital twin might work in practice. The goal was to explain how these systems might provide value in practice so public sector entities all over Europe would know how digital twins could provide value and understand what kind of services they need to procure.

 “We want to make sure that the private sector can deliver those digital twins that can provide value,” he said.

Data sharing is crucial

They realized that the data sharing aspect stood out as a critical requirement for scaling these local digital twins to support more types of decisions and new use cases for businesses and citizens. The researchers began investigating how an emerging concept called data spaces might provide the foundation for this new vision. 

The core ideas behind data spaces have been around for several decades. In 2007, the European Union issued the INSPIRE Directive, which mandated national and local governments to develop geographic information system (GIS) data sharing programs. But it left each government to decide the technology and rules for doing this. This led to over 150,000 data sets. 

At the same time, commercial data sets such as Google maps, various satellite services and IoT sensor data streams started to grow in size and utility. People realized they could create new value by combining these data sets. But technical, legal and business agreements had to be freshly negotiated for each point-to-point integration. 

And things have grown more complicated for businesses with a complex regulatory landscape designed to protect individuals, including the Digital Services Act (DSA), the Digital Markets Act (DMA), Digital Governance Act (DGA) and the General Data Protection Regulation (GDPR). “There is a big concern that this regulation is needed, but also that it could hinder the European competitive position compared to the US and China,” Coenen said. 

Promoting data flows

In 2017, the International Data Spaces Association was founded to help create a framework to automate data sharing, negotiations and compliance at scale. It spells out security standards, stipulated control and enforcement for data usage and specific rules for data traceability. The group has been evolving the core framework and released the latest reference architecture model, IDSA RAM 3.0, in 2019. 

“The essence of a data space is to enable data flows because everyone knows what the agreements are up front,” Coenen said.

In parallel, another collaboration emerged to focus on technical cloud interoperability in Europe called GAIA-X in 2019 to design the next generation of a federated European data infrastructure. Coenen said both GAIA-X and IDSA now play a vital role in the rollout of data spaces. 

Motivating adoption data sharing standards

Both IDSA and GAIA-X are helping to define the technical standards for sharing data across clouds. This will require a lot of investment to bootstrap these data spaces to ensure that there are enough participants to make sure it grows and works. Coenen also believes it is essential to inspire business leaders to create new business models that take advantage of and grow data spaces. 

But this requires some guarantees and contracts from participants. “If you build a critical infrastructure application on top of my data, you don’t want to risk the fact that I will bail out on you at some point,” Coenen said. “It is not complex technically, but it is highly complex legally. “

Another important consideration is protecting data sovereignty. Individuals may be pleased to share personal data if it leads to better government services, but may be reluctant to give it away to private companies. The Government of Flanders, a Belgium region, plans to give each citizen a personal data pod on Solid Inrupt, a decentralized data sharing platform. The new platform promises to make it easier for citizens to share data while maintaining ongoing control over how it is used. 

An ML model economy

Cities are just starting to bring a lot of different data sources into digital twins, but municipal planners are all trying to figure out how this data can address their concerns. “We are advocating there needs to be a marketplace of ML models to help you understand your data,” Coenen said.

The technical infrastructure will help push these models to parts of the IT infrastructure where they can provide the most value. Distributed models could run on edge servers at traffic intersections to analyze and aggregate local traffic, pollution and noise data. Other models for aggregating data across the city will need to run on high-end GPUs in the data center. “Finding ways to push models to the correct executable to run on the appropriate infrastructure is significant,” Coenen said. 

Different models could help predict things like traffic flow, air quality and noise and they could be linked together. “Dataspaces will enable this market for digital twin machine learning models to talk to each other,” Coenen said.