Fifty years ago, NASA kept full-scale models of its space capsules close to hand to help it diagnose problems with real capsules out in space, and to help it come up with fixes to problems that it could communicate to the faraway astronauts.
The idea of a digital twin is something very similar. However, instead of building a physical mock-up of a space capsule, or any other physical object for that matter, a digital twin involves making an accurate digital simulation of the object which exists only as computer code.
The Function of Digital Twins
In many ways a digital twin is similar to a virtual machine in that it is a digital entity, which is designed to mimic the workings of a real object. But a virtual machine is designed to be used as an alternative to a real, physical computer. By contrast, a digital twin is designed to be tested and experimented with, but the ultimate aim is to use any insights gained from the digital twin on the digital twin’s physical counterpart.
A good example of this is the use of a digital twin of a car and digital crash test dummies. Car designers can run a digital twin of a car and one or more dummies through many different types of crash scenarios and use the data from these crashes with the ultimate aim of making safety improvements to the real car’s design.
Read more about Digital Twins.
Digital Twins and IoT
Creating accurate digital twins of space capsules, cars, and crash test dummies is an incredibly complex business involving researching the physics of these items and then developing a precise mathematical model that describes them and their behavior.
But the good news for those involved in IoT is that the “things” in question are often relatively straightforward sensors, and these can be orders of magnitude more simple to model mathematically than something as complex as a space capsule or a car.
That means that creating digital twins of many types of IoT devices is relatively straightforward, quick, and crucially, inexpensive.
It’s also the case that many IoT environments consist of large numbers of simple identical devices, such as temperature or humidity sensors in containers, or GPS units in vehicles. That means that once a digital twin of a device has been created, simulations of large numbers of these devices working together can be created simply by making multiple copies of the digital twin and feeding them with data. This can be “artificial” data, or data that is received by existing physical devices.
The implications of this for large scale IoT deployment and management are huge, as we shall see.
Also read: SD-WAN is Important for an IoT and AI Future
Multiple IoT Applications
At the very start of an IoT initiative, digital twins can be used as device prototypes to help fine-tune the precise design of the device itself, as well as its firmware, encryption systems, and other software.
Once this process is complete, digital twins can be used to help optimize the deployment of devices by testing how many devices are needed in practice, where they should be positioned, and how they should be connected through various networks to data collection hubs.
Testing Updates and Changes
Once large numbers of devices are deployed, digital twins can also be used to test firmware and other software patches and updates before they are sent out over the air to their physical counterparts. This can be particularly useful when changes are made to the way that devices interact with each other, as large scale simulations allow developers to see what the results will be before the patches and updates are deployed en masse.
Digital twins can also be used to help design and manage changes to network topography, and even where data is collected. For example, an organization may be collecting data from its IoT devices on servers in its own data center, but as the IoT network expands it may decide that data needs to be sent to the cloud for collection.
In this sort of case, digital twins can help predict when a changeover would be necessary, how performance might be impacted either positively or negatively, and what scale of cloud resources would be necessary to get a required level of data collection and processing performance.
This sort of use of digital twins in IoT networks is what might be called “predictive twins”. Rather than using real data, a network of digital twins used as predictive twins could also be used to test the impact of different types of data flows, increased data traffic, and many other situations, to see what the impact on the IoT network would be, and what changes might be needed in the future.
Digital twins could also be used as predictive twins in the sense that they could also be used to predict when maintenance or replacement of the physical counterparts might be necessary given many different usage scenarios.
The Value of Digital Twins
Digital twins are uniquely suited to IoT deployments because of the relative simplicity of IoT devices and the fact that digital twins can be replicated at little or no cost. Recognizing this, Gartner predicts that digital twins will exist for “billions of things” in the near future.
The possible benefits of digital twins to organizations with IoT deployments are staggering, said David Cearley, a Gartner vice president. “Potentially billions of dollars of savings in maintenance repair and operation (MRO) and optimized IoT asset performance are on the table,” he concluded.