3 Considerations When Using Digital Twins
by Brad Johnson, on Jun 4, 2018 2:16:10 PM
STATEFUL SERVICES, THE DISTRIBUTED ACTOR MODEL & MACHINE LEARNING
I’ve been attending a few technology conferences recently, and it seems like there’s a renewed buzz around digital twins. If you’re not familiar with the concept, a digital twin is a software representation of a physical object, such as a machine, sensor, or real-world process. Digital twins can simplify how real-world objects are managed and coordinated within software applications. Paired with machine learning, digital twins can enable predictive maintenance for industrial equipment, or efficiently monitor individual assets flowing throughout a complex supply chain. Though not a new concept, the concept of digital twins has been around for years, with increased focus on IoT technologies and innovations in edge computing, the digital twin is poised for a breakout.
By the year 2021, Ericsson anticipates that there will be 28 billion connected devices deployed. These devices will generate data, and the hidden insights, to reduce costs and optimize business processes. But to capture that value, we’re going to need a simple, repeatable way to manage and interact with the unbound numbers of connected devices. Digital Twins provide an efficient way of doing just that.
Here are 3 things to consider before using digital twins for your next project:
- Stateless (REST) vs. Stateful Services
Most applications today are built using the stateless REST model, where application state information is stored in a local database and queried as needed. Digital twins often involve real-world machines and devices which continuously require application context in order to power local control, inform monitoring applications, and to predict maintenance needs. In these kinds of applications, it makes sense to build using stateful services. Stateful services are always “aware” of application context, even at the edge, and can enable local processes in real-time, without waiting for a database.
- Distributed Actor Model
Today, digital twins typically refer to a generic model of all like objects within a system. These models are constructed in the cloud, and then either run in the cloud or deployed to edge datacenters to run locally. However, a distributed actor model enables unique digital twins for each object in a system. This allows for the creation of thousands (or even millions!) of digital twins which can each accurately observe and predict for their own unique application context. For example, SWIM.AI created a unique actor, or digital twin, for each intersection in the SWIM.AI Smart City demo application.
- Machine Learning
Digital twins and machine learning make for a potent pairing. Using deep neural networks or other machine learning techniques, digital twins enable learning in parallel on distributed data streams. These machine learning models can generate predictions for each individual object in a system, and be aggregated to make higher level predictions about system-wide behavior. Digital twins provide an efficient way to learn on real-time data streams at the edge, when data volumes may make it infeasible to learn in the cloud.
Learn how SWIM EDX uses digital twins to deliver real-time operations insights from the dark data generated by IoT applications and other connected systems.