TMU researchers rethink digital twin design with a simpler, systematized method

Digital twins, virtual replicas of physical systems that continuously exchange information with their counterparts, offer the potential for huge real-world impact.

However, designing and building a digital twin, along with the systems that support it, is complicated, time-consuming and resource-intensive work. The process demands collaboration between subject-matter experts, software developers and other specialists who must overcome platform inflexibilities to knit multiple systems together and create a cohesive virtual entity. The end product must easily and reliably communicate with whatever asset it’s twinned with, ensuring consistency between the physical and digital environments.

In light of these challenges and the limits they place on digital twin development and deployment, Toronto Metropolitan University (TMU) computer science professor Sadaf Mustafiz is rethinking how digital twins are designed and built. She’s leading an interdisciplinary team of graduate students whose research seeks to simplify and systematize the systems design process behind digital twin creation to make it faster, more accessible and more reliable.

“Digital twins are often difficult to build and even more difficult to maintain,” professor Mustafiz said. “They tend to be fragile in the face of system changes, and don't adapt well as systems evolve. We want to move beyond bespoke, one-off prototypes to engineering-grade digital twin technology that is repeatable and reusable. By starting from a common design, we can systematically build both physical and digital twins, making them more scalable and reliable.”

Simplifying the digital twin design process

To reach that goal, professor Mustafiz and her team didn’t start by writing the code for an entirely new digital twin system. Instead, they relied on a model-driven engineering approach based on statecharts. Also known as multi-state machines, statecharts are widely used in computer science for behavioural modelling, simulation and code generation.

“In simple terms, statecharts are like flowcharts that are functional and executable,” explained team member Sahil Salma, who recently completed his master’s degree in computer science at TMU. “We use them to condense the complex behaviour of the physical system into computer-readable models. The idea is that instead of building out the code from scratch, we have the design of the entire system figured out as a model. Then we can use the design to automatically generate the code.”

The team used statecharts to design a smart home application with centralized control of a smart lighting system, garage door opener and fire alarm. Their digital twin communicated effectively with a real-life working model built in a dollhouse furnished with items borrowed from professor Mustafiz’s daughter.

Sahil Salma was the lead author on the team’s paper, Engineering Digital Twins with Statecharts: A Smart Home Application, which he presented last fall at the second International Conference on Engineering Digital Twins. In addition to Salma and professor Mustafiz, the project got contributions from computer science graduate student Zenan Zha and research assistant Protik Mukherjee, whose background is in electrical and computer engineering.

Ensuring consistency between physical and digital assets

The behaviour of digital twin systems – how they act, react and coordinate – is of particular interest to the research team, and with good reason.

“We treat behaviour as the single source of truth for both the twins,” professor Mustafiz explained. “If you don't have behavioural consistency, then any simulations, predictions or recommendations generated by the digital twin can’t be trusted.”

Professor Mustafiz and her team are leveraging what they learned from the smart home project to examine how their ideas can be applied to Controlled Environment Agriculture (CEA), such as greenhouses or vertical farms.

Read the paper, "Engineering Digital Twins with Statecharts: A Smart Home Application," on IEEE.