This article was edited by SPRITE+ Research Associate Dmitry Dereshev, with responses and edits from IOTICS Co-founder and Inventor Mark Wharton and IOTICS Head of Customer Success Sophie Peachey.
How would you describe your respective jobs?
I am Mark Wharton. I am co-founder and inventor of IOTICS. We have seven patents, and my name is one of the joint names on all seven.
I am Sophie Peachey. I am head of customer success. I am the voice of the customer within IOTICS making sure that they have got what they need from our product and translating any needs back into the business. In Iotics I stand up for the customer.
What are digital twins?
Mark: A digital twin is a virtual representation of a real-world asset: it could be a sensor, an engine, a share price, or it could be a person. You can make a digital twin out of anything that can be described by a set of metadata, e.g., what it is, where it is, and whatever you else you want to say about it. Digital twins do two things: they publish data about what is going on in the real world via a feed, and they can also respond to commands or data coming from other digital twins.
Some people think of a 3D model of a building as a digital twin, but if it does not have any data other than the structural data of the building, then it is not much use. However, if you were to walk into a virtual version of that building and it tells you what the temperature is, how much power it is consuming, how many people are in it, what the fire safety regulations are saying, that kind of thing – then it is a digital twin.
Sophie: From a client’s perspective a digital twin is the way to observe the assets that they sell. For example, Rolls-Royce worked with us to create a digital twin of a power-generating unit, so that if their client had an issue with that unit, they could get all the necessary information to the person who is responsible for dealing with that issue. So, a maintenance engineer can get all the information about that asset brought to them, instead of having to query several separate systems to find out e.g., when, and how that asset was manufactured, what the conditions were when the issue happened, etc.
The way that can evolve for customers is that we can also work with ecosystems of digital twins to create context. Following the same Rolls-Royce example, the power-generating unit we have created the digital twin for was powering a train. With that you can start to bring in information about the train, the depot where the train is being maintained, the weather, the pollen levels that the train is passing through, etc. With that you have got a greatly enriched picture about the environment in which your asset is performing. The benefits accrue depending on how much a trusted ecosystem grows up and around the digital twins that you are creating.
Are digital twins difficult to create?
Mark: The hardest thing is probably the descriptive part of it and the modelling. How do you describe your digital twin unambiguously to somebody who is not in your domain? We have this joke about sending goldfish into battle because somebody said: “I need a tank” and they got a fish tank rather than a military tank. Solving these interoperability problems is always going to be difficult.
Temperature sensors, for example, must have the following metadata: the manufacturer, the location, and the feed of temperatures over time. You can make simple models like that but try and make a simple model of a railway station – that is trickier. This is where programming skills are involved. If it is straightforward, there is an API we create, and then you give your digital twins the necessary properties.
How do digital trust, identity, privacy, and security work in digital twins?
Mark: Trust is built around a peer-to-peer network based on a distributed hash table. The node in the hash table is what we call a space. A space is a bunch of servers that allow a person or an organization to create a set of digital twins that they keep for themselves. They can choose to make those digital twins publicly available, in which case they will be searchable by other people in the network, or private, in which case they are only visible to the owner. So, it is a network of disconnected spaces that interoperate as and when they need to. The privacy piece is which specific information you choose to share, and which you keep private.
The Iotics vision of the digital twin ecosystems: private and public data linked between many providers with various levels of trust.
Sophie: We support the evolution of trust. For example, with Rolls Royce, the power-generating unit was going on a train, so the first conversations that we had were with Hitachi who were the manufacturer of the train, with Network Rail because the trains were running on their lines, and with the First Group and Great Western who were running the service that these trains were operating on. There is a wider ecosystem that this one component, by virtue of being on a train, is a part off, and it might have information that could be of value.
The most useful information that the power-generating unit had about itself from Hitachi's perspective, was its location. For them it was useful to know where the trains were and whether they were on schedule. That informed their planned maintenance activities for that evening. With this, you have suddenly established a level of trust where Rolls Royce is willing to share this one bit of information about where its asset is and that in turn allows Hitachi to look at the information it has got from its digital twin of the train or the digital twin of the depot, and think: “What might be useful to start sharing back into the ecosystem?” All these parties are currently doing a very good job of managing their own little bit of the world and are now starting to be able to improve the way their fellow ecosystem partners can work just by sharing little bits of data. That increases trust.
We can also enable ecosystems where there is not necessarily much trust. If you think of a wider rail ecosystem with lots of different train operating companies which are competing with each other, they are not going to share all their information with one another. We can enable a marketplace which competitors can use to share or trade information. If anybody is interested in getting that information, they would go to the marketplace to get it rather than the spaces of the individual organisations. So, you have got this ability to create quite different ways of working compared to what people have been able to do before outside of a collaborative research environment, and I think that is exciting.
Mark: We have also done some work with MULTOS. It is about proving cryptographically that this sensor is represented by this digital twin. When you are in a virtualized world, something could come along and claim to be that sensor. On the Internet, nobody knows you are a dog. But with the MULTOS chip and the way that they do identity, you can prove that this chip on this device has signed and encrypted this data to create the identity for this digital twin. Then you can prove that the digital twin and the device are the same thing.
On the Internet, nobody knows you are a dog.
We use the W3C standard for decentralized ID to allow people to create their own identities based on certain rules. You also get a document which is a JSON-LD document about that identity and their crypto proofs. That's a very valuable jump from what we used to have when we used to give digital twins a globally unique ID. You cannot have a decentralized system if someone central holds on to it.
How would I know if an asset owner is who they claim to be?
Mark: That is a good question. This would probably have to be done out of band, e.g., by Hitachi directly communicating with Rolls Royce to confirm which digital twins they own. It is difficult to do within the digital twin space, and it is not something that Iotics can do at an individual level. Perhaps putting a certificate within a digital twin that only an owner of that asset can own would prove that the space is indeed owned by the that specific owner.
If I want to find a digital twin – where do I go?
Mark: If you come to Iotics, you will have a digital twin space where you can say: “Hello space, could you ask the other spaces if they have got any of these kinds of things?” The point of Iotics is to make an abstract version of this with a unified way of doing these searches and twin descriptions. So, if you are interested in, say, rainfall in Swindon it will find relevant public rainfall monitoring devices for you, regardless of who owns them. Air quality is another great example. There are no consistent air quality measurements across the UK. Electric vehicle chargers are another example. Think of something that lots of people do that is not integrated, and that is the point of Iotics – to bring that information together.
What kind of information do organizations publish in their digital twins?
Sophie: This is evolving. In the first instance organisations are comfortable only sharing information internally, but it is only when they start to look beyond their own walls, that they discover new use cases. It is more about what needs to be shared to help in a specific situation.
So far, the ecosystems that we are supporting have not included any financial transactions. Companies are working within an ecosystem where they all benefit by sharing data. We are yet to see a marketplace where somebody is not giving anything back themselves but taking value from the data that is being shared. I think we are seeing the requirements for that forming with things like the railway data marketplace. There is an expectation that a whole service level will start to grow on top of data that other organisations are sharing. And that is when you might start to see financial transactions.
What are the challenges in the field of digital twins?
Mark: The first one is security. I would say people are paranoid about sharing. They think that sharing their data is like opening the floodgates to the whole of their datasets and everything. Getting them to realize that they can use their digital twin as a medium for sharing as much or as little as they need is our biggest challenge.
Sophie: Digital twins are often wrapped up in the creation of a new business model with different financial and legal adjustments that the companies have never done before. It is an entire mindset change, and that is incredibly challenging for organisations. While yes, the technology is brilliant and it offers massive opportunity, there is this hidden pain of trying to adopt it from the perspective of organizational change. It takes support from people who are skilled at business transition and digital transformation to help enable this kind of move.
What problems, if solved, would significantly advance the field of digital twins?
Mark: One is semantic interoperability: say I have got a BMW car and you have got an Audi car. How would they talk to each other? To the highway? To a smart city which they have never visited before? That kind of interoperability across domains, and across interpretations is a hard problem.
Another one is provenance. For example, if a digital twin claims to be a temperature sensor, and it says that its temperature is in Celsius, and it is 26 – how can I trust that that twin is indeed a temperature sensor, the temperature is indeed in Celsius, and its current reading is indeed 26? What are the error bars on 26? And if I am going to use that temperature in an aggregate across several sensors, can I trust the algorithm that does the aggregation? Those are the tough challenges in my view.
Sophie: For me it is always the adoption issues of understanding risk and going through that mindset change. We have a generic capability that can be applied in so many different situations and trying to identify replicable value is hard. We have worked with people living with the early stages of dementia and how to help them remain independent for longer; we have worked with engines; and we have worked on building sites. It is hard to pick out a common theme of value other than “it has been really, really useful to share data”. If I talked to somebody in aerospace about the benefits that the people on the project with the early stages of dementia experienced, it is a different way of expressing it, and it does not fit into their world.
It is also quite hard to monetize that value. The first use cases we tend to hit are operational efficiencies: the operational efficiencies with train engines; a more effective maintenance on the building site. You are not making money, but you are potentially saving money just by doing things a bit better and making life easier for everybody. It is hard to place value on that because you really need to understand what the cost is now, what needs to change, and whether your change made an improvement.
The second set of use cases tend to be the revenue generating ones, creating a new data service out of sharing data. In this case you are creating something brand new and there is no benchmark, so all you can say is: “we have now got a new revenue stream and it is bringing in X amount of money”, but there is no real indicator as to whether we could have made that money by doing something different because it is such a new approach.
What kind of community around digital twins do you envision?
Sophie: We are building communities that are very customer-focused and very business-ecosystem focused. For example, in the rail ecosystem you have organizations that are trying to make experience better for the customer: be that the passenger or the freight customer. How do you make the entire UK rail network more effective? – Let us start with our own little bit of it and see if we can work with our partners to make our little bit better and then gradually expand that.
The community building has been very business-focused because we are having to prove the value of creating and using digital twins, and make sure that organisations realize that this is worth their time. It is not always an easy conversation with customers and the customers’ customers, because digital twins hold up a mirror to you about how well you are doing things yourself, and it can be a bit embarrassing if somebody else has a quick look in that mirror as well.
Mark: We have a demo site where you can get an account on request that has got access to all sorts of open data on digital twins like Transport for London, bike stations, air quality, river levels, rainfall, and so on. They are all in the same platform, so you do not have to collect that data from disparate sources.
Sophie: We also have our data for good program. We are actively looking for communities including academics to collaborate. A good example is as a living lab, where they are actively looking to make assets within the lab available for other people to use. Anybody wanting to use a living lab can now engage with the digital twin of it and all the data sources from within the living lab are easily available to them. They can use the living lab but leave no footprint. They could turn up to a lab and use it if they needed to do physical work, but they could do a lot virtually.
We are looking for interesting collaborations, particularly those that are focused on something that is going to benefit humanity and the world. We are sponsoring these initiatives by providing our technology for free.
Final remarks
Mark: There is a tendency for everybody to try and invent digital twin infrastructure themselves. People think: “How hard can that be? We will go and make one.” At Iotics we have spent eight years doing it and it is clearly quite hard, but now instead of reinventing the wheel, people can come to us. We would like to improve Iotics, and we will give this infrastructure away for free, so that people can play with it, solve their problems and come back to us if they want to pay for it in the future. We provide a secure and well-developed infrastructure for digital twins. If you have got a sensor and you want to make it publicly available on the Internet, then just come to Iotics and we will do it for you. You do not have to build all the infrastructure that allows that to happen from scratch.
Sophie: We let people be experts at what they are expert in. We are very, very collaborative and we would much rather that people did not have to worry about connecting data sources or other related infrastructure issues. Let us be your experts at semantic interoperability and secure live data and use your time to be an expert at what you want to do.