CCC Responds to the NITRD Request for Information on Digital Twins Research and Development
By Catherine Gill, Program Associate, CCC
Envisioning a future of digital twins leads to near limitless possibilities. Researchers predict digital twins will become increasingly personalized, with every individual having access to a digital twin of their own body. Imagine receiving real time updates from your phone to monitor your health conditions and predict health crises before they happen, or athletes receiving instant updates on their training regimens and recommendations for improving. This future is possible, however significant further research is necessary.
In addition, researchers and stakeholders need to be realistic about the current capabilities of these models, as well as informed about the future of development. That is why the Networking and Information Technology Research and Development (NITRD) National Coordination Office (NCO) is developing a National Digital Twins R&D Strategic Plan, to guide government investment in digital twins research and fast track development of this technology to address national priorities. The NITRD NCO released a Request for Information in late July to help inform the development of this strategic plan, and the CCC submitted a response.
In our response, we emphasized several key considerations for developing a strategic plan, one of those being that digital twins should not be treated as a novel invention. Digital models have been used since the 1950s to predict weather patterns and model the efficacy of different airplane designs, to name a few applications. Recent advancements in model capability, networking, and sensor development have allowed digital models to become more accurate, but we’ve been using very similar models for around 70 years. Digital twins themselves have even been in use since the 1960s, such as those employed by NASA to simulate spacecraft in space exploration missions.
With all of the recent innovation in generative AI, especially in image and text generation, it is easy to forget that AI models have limits. However, in practice, digital twin models do not store every bit of data or learned information. They also do not always store data at the highest granularity, because doing all of this is expensive in terms of energy consumption and necessary infrastructure. In our response, the CCC emphasized the importance of not over estimating the capabilities of these models.
Cybersecurity concerns are one of the largest considerations when implementing digital twins, because they create a new surface of attack for adversaries to exploit. To prevent unauthorized access, developers need to secure every endpoint of the system. Cryptographic protocols and algorithms can be used to prevent malicious corruption of a digital twin via the physical system. We also stressed that digital twins must be securely developed from the beginning, not as an afterthought.
Finally, we underscored the importance of co-design and establishing interdisciplinary teams to develop these models. For these models to be beneficial to end users and primary stakeholders, they must be developed with these users in mind, as well as the key features needed for the model’s functionality. Development teams should also be composed of experts across disciplines to ensure the models are implemented accurately. For example, a digital twin of a bridge should consult the engineers who built the physical bridge, to ensure the sensors in the physical bridge are accurately represented in the digital twin. City planners and environmental experts may also be useful to consult, to have a better understanding of how many people typically use the bridge on a daily basis and to gather what simulations may be beneficial to run on the digital twin, such as hurricane or extreme winter storm simulations.
This CCC response was written by David Danks (University of California, San Diego), Catherine Gill (Computing Community Consortium), Chandra Krintz (University of California, Santa Barbara), Brian LaMacchia (Farcaster Consulting Group), Daniel Lopresti (Lehigh University), Mary Lou Maher (Computing Community Consortium), and Pamela Wisniewski (Vanderbilt University). Read the CCC’s full RFI response here.
