Traveling between Digital Twins
One of the most prominent truth about the digital twin, it becomes every day a part of our life, even you don’t feel it. Digital Twin(DT) is a way of copying a thing from its environment to another environment to understand it in different perspectives with the aim of increasing or optimizing its performance. Its main distinguishing point from the Digital Shadow (DS), briefly, it is more than monitoring. If you receive data from a device or entity and only monitor it, this is called as Digital Shadow (DS), if you consume the collected data to execute some operations and interact with the entity, then this is called Digital Twin (DT). We can think of DS as a sublayer of DT.
After a short terms introduction, my intention is to direct your focus to another point in the DT. I am pretty sure you remember the film series of “Back to the future”, in which the professor and the main film character were travelling from now to the future or from the future to the past. Any change that they made in the past creates a totally new path in the future. Indeed, the same thing could happen also in DT world, just in a small scale.
Assuming that we have a production line that is composed of a number of sensors. All these sensors generate continuously data, called time-series data, and they are stored in the databases. All these sensors may have an effect on each other or not, this can be only figured out after a thorough data analysis. In case we intend to look at the past data of the production line, we select a time interval in the past to understand what happened at that interval. What we did for the production line was the time travel for the device data. Even the observation of the data is partial, since we cannot store all data in atomic level, this time travel may not reflect the all details of the production line, nevertheless, our data sampling will be enough for some use cases. Now we proceed one step further and bind with the DT, which can create a number of simulation cases relying on the data and the entity metadata, and it can end up with the possibilities world for this entity. In the following figure, the historical data are collected from an entity, and based on that, DT tools can generate multiple simulated entities, and they can result in distinguished targets.
This case can be also connected with many science fiction movies, in which the characters try different approaches for the same scenario to find the best solution for the existing problem.
In the figure above, only a single figure is depicted. If multiple entities and their connections are considered, the complexity level of the DT increases and the variety of the simulations, predictions or optimization approaches will also augment dramatically. The reason is, all simulated entities of the real entities within a production line will be chained subsequently. Any change in the simulated entities would create an individual path or case for the production line. A single change can lead to the generation of a huge tree change in the future path. When the environment conditions should be also involved, then the problem surface increases exponentially. In the end, the existence of the historical data and simulated data created on the historical data enable us to look at the past data, and its potential future data. For this reason, we can claim there is a similarity between the “back to the future” film (and of course some other science fiction films) and digital twin.
Production line scenario is one of the small scaled example that can be imaged for the digital twin. As long as there is a computation resource limitation, it is difficult to realize complex scenarios. What we do for the time being is instead of collecting and processing all use-case data, we extract only the important part from it considering our requirements, which targets mostly an energy efficient, less time and resource consuming solutions.
DT would open new ways of thinking in the coming years and broaden our minds, however, there are still many challenges that have to be overcome to create a well working DT system.
I want to conclude this text with a question to just investigate or even understand whether there is already a DT case in the universe. For instance, does universe store also its data, if yes, how? What we can see, the old signals that are far from us, and probably not more than that in the light of our current knowledge. Assuming just a second all our events are stored, how can we visit one of the past digital world copy as in the “back to the future” film. To better visualize, I put a figure above to show what I actually imagine. A similar question would be “Does the universe have long term memory?”.