Evolving Digital Twins
Digital twins are like a social media profiles for things
- Birth date
- Network orientation
- Certifications
- Contact info
- Connections
Every device is a spam bot
Scale of data generated by devices puts Twitter to shame A Twitter addict might tweet a couple times an hour It’s not uncommon for devices to “tweet” several times a second
It’s difficult to communicate with devices
Connections are unreliable Devices speak varied and esoteric languages Security needs to be strictly enforced
Easier for applications to communicate with digital twins instead
The digital twin is always available on the cloud, even if the device is disconnected.
The digital twin ensures that the configuration of the device is kept consistent with the parameters prescribed to the twin.
Digital twins have more compute at their disposal than embedded devices
The device has some dinky cost and power-optimized ARM chip Digital twins run on supercompute clusters on the cloud
Digital twins can act on behalf of the underlying device
Like an agent And sometimes like a puppeteer of the device.
Digital twins can coordinate with other digital twins
Even if the underlying devices are unable to directly communicate.
Digital twins protect their underlying devices
Digital twins gate access to the device, and to sensitive information about the device, thereby guarding its security and privacy.
Digital twins can pretend to be the device as part of a simulation
So you can try out new behaviors virtually before risking a real-world change.
Digital twins can pretend to be a device before the device even exists
This is called model-based design. It’s used to simulate factory configurations or cell network deployments before ordering—or in some cases even designing—the hardware.
Major plot twist: sometimes there’s never a device
Digital twins can be used to model things that aren’t devices at all.
You can create digital twins of customers.
Digital twins of product lines.
Digital twins of whole stores.
Digital twins of cardboard boxes (a.k.a. packages)
Digital twins of wavelengths of light (i.e. RF spectrum)
Digital twins of regions of space (map tiles)
What makes digital twins different from other software models?
The key is that digital twins are in some way kept coherent with a corresponding thing that exists outside of the digital twin model
Digital twins are the ideal destination for streaming data
Streaming data represents what’s actually happening in the real-world right now—exactly what digital twins are designed to model.
Digital twins are the ideal place to join data streams
Route all messages—from all sources—about a particular thing to the digital twin of that thing.
Digital twins are the ideal place to perform real-time compute
Digital twins maintain real-time state and context.
Digital twins can execute real-tine business logic as streaming data flows through the twin
Digital twins are the ideal foundation for streaming applications
Applications are built on entity-relational models. Digital twins are identifiable entities with definable relationships.