CEI‑Spheres and the hourglass model

This talk discusses the CI-Sphere project, a European Commission-funded initiative aimed at fostering collaboration for open source software strategies. The speaker emphasizes the importance of developing a robust ecosystem through the hourglass model, which bridges the gap between supply and demand in technology innovation. The project aims to establish large-scale pilot cooperation while promoting data-driven understanding and privacy-preserving edge ecosystems. Various experts from different organizations share insights on the operationalization of these software strategies, focusing on real-world applications and market validation. The presentation also highlights the need for interoperability and standardization, particularly in the context of energy systems, mobility, and industrial productivity.

[music] >> Thank you for having us. Thank you for enabling us actually to have this very interesting session. Um so let's start. Just to give you a bit of introduction about the session. So we start now. We have welcome and then we will see objectives from Rolf from the commission. Then we will give a bit of introduction about one of the EU projects called CI-Sphere. Um and

the hourglass model that came out of the there and then of course what are the links with the open source and with the research. And then we start with the individual, let's say, presentations of the series of experts that you see on a screen. Some of them are also related to um some of the EU projects, not in general about those projects, but how these projects can

be leveraged um in the sense of the operationalizing of the open source software strategy. And then we will continue with the panel discussion after the lunch and then we end up with the Q&A with the audience. Now on the speakers just we will not have enough time to get into details about each of the colleagues, but here is the here is the list. So Rolf from the

commission, uh Ignacio Nacho from the Universitat Politècnica de València and also from the OCI project, Giorgos from the Ubitricity and also Co-pilots, so two big um large-scale pilot projects related to the also project called Uh Lucian from RTE. He will also talk about um links, let's say, with the Linux Foundation. Antonio who is also part of the CI-Sphere OCI project, Giovanni, member of many other, let's say,

new projects that are coming like E3C, right? And include. Just as an example, Manrique from our host Eclipse Foundation. Um and then myself from BlueSpecs and also from couple of already mentioned projects. So this is this is how it goes usually. Okay, so just to start with a very brief um introduction about CI-Sphere project, which is one of the coordinating support actions financed by the European Commission.

So what about the projects? You see basically here what are the five kind of key um I would say things to develop out of the project. So establish solid ground for inter large-scale pilot cooperation, large-scale pilots OCI Co-pilot that you um see here and you will hear from them in the next presentation, so we will not go into that now. Develop extensive data-driven understanding of the European

CI market. That's quite important to see where the sort of common architectures, common approaches, synergies that can be built. Support the privacy-preserving open edge ecosystem. Building a spheres, so spheres we will explain a bit later what this is. And technology backbone toolkit that Antonio will talk about also. And then the curate also the stakeholder CI stakeholder ecosystem. Now on the main activities definitely is the spheres like

the name of the project calls, which is basically to map um use cases from the two already mentioned big large-scale pilot projects into the spheres. We will have we have three spheres. We will show you in the next slide which one are there. And then to identify the key challenges for these spheres and then basically how this can deployed in the market and what is the market,

I would say, reaction to this. Building the use case catalog and then to see also can we make some sort of the alignment, let's say, with Eurostat. Antonio will talk about this later. Open calls which are run by the two big LSPs and then what are also the synergies that can come come out of it and then organize hackathons. Um we will basically co-organize one hackathon with

OCI Co-pilot and Eclipse end of June in Cagliari and then hopefully we will also be able to provide inputs to the Eclipse SDV hackathon in October. On the spheres you see here there are three that we identified through basically assessing the use cases of the two large-scale pilot projects. Energy and infrastructure resilience, you see 17 use cases as a primary, 11 as a secondary. What we mean

by this is that of course in a lot of use cases you have a sort of overlap or let's say joint um let's say joint domains in each of them. So for example, if you talk about vehicle to grid, obviously that's about mobility, but since you have grid, obviously this is about energy as well, Then mobility logistics and a software-defined vehicle. And then industrial optimization and productivity.

And you can see that this is almost almost equal distribution of use cases. That's not by purpose, that simply came out like that. But I think it's quite nice that it's in a way balanced, not not that like one area has, I don't know, 30 and the other area has five. So this also shows actually where the interest goes. And at the end ultimately and hopefully these

spheres should show also sort of the test with the market, sort of the market validation, how this sits and feels with the market because the both projects should be innovation actions, which means very close to the market. So these are not theoretical um research lab activities. These are actually activities that are very close to the market and hopefully we can help to see actually what what could

be the possible market uptake of these use cases. Uh we also have a three task forces that are supporting actually the spheres and the whole work together. Task force one, which is about the open source architecture standardization led by Antonio. Task force two, which is a market uptake industry engagement that we will take a um we will talk a bit a bit later about that. So go-to-market

pathways, tech briefs, open calls product. And then task force three, which is a stakeholder forum outreach. Of course there is a link to other um technology platforms, other initiatives. This is not like isolated case working around its own, let's say, X, but this is a part of the larger ecosystem and we want to actually um get engaged with them and also see how this works together with

them. And then we try to see also how we could align with the initiatives like the Eurostat. So we try to do um what we call the hourglass model, which Antonio will explain exactly after slide. And we try to see also then how uh so so every use case from both projects is mapped against the Euro hourglass and then we try to see, okay, what could be

the possible links to the Eurostat initiative. This is just one example from the Co-pilot one industry cluster mapping. And Antonio will give us a bit more details, but why this is important? Because this is a sort of the leeway where we also start to talk about open source components as well, right? So that was for the introduction. Thank you for your attention. And then I think we

can continue with slides from Antonio. Um I think it should be Oh yes, you're right. So first of all, let's switch to Rolf and then we will come back. >> right, right. Rolf. Give us give us Okay. So it should be this one, right? You can come here. All right. Thanks so much. Um Good morning, everyone. Very good to to be here. I'm in the unit called

now intelligent mobility, energy and IoT. Our main stakes in software-defined vehicles on the one hand, but also in the transition of the energy systems. And looking at what we call on the research. And it was mentioned on research um we spend more than 300 and now the new program more than 500 million on what we call cloud edge IoT continuum. So it's continuum in the area of

uh um mentioned the energy sphere, you mentioned the mobility sphere, but also now the telco cloud edge. It's continuum there, but as I said, research has also double sword. It also has this angle of policy support. And policy support means yes, I think we talk about um key elements like um the sovereign package. I think people talked about that yesterday. Uh but also they supported a conference

the policies on smart by direction charging last November. Um to be to be visible. And you also heard now in the panel just before on the CRA, which is a burning issue for many of the companies developing what we called software-defined open source programmable infrastructure. So I think our main stake in the unit is to look at the industrial software stack for future infrastructures. Yes, and to

give you one example and I think what drives us at the moment what commission policies is artificial intelligence. I think it's all over. And one example is the energy system is one of the most complex systems ever built, one of the most largest systems, distributed systems. And it also has to go and then transition towards decentralized systems and small software-based systems. So here we have a real

challenge over there to to see what's happening if we introduce intelligence and automation. How to oversee the whole system, but the advantage of AI is it can deal with complexity and increasingly uncertainty about the user behavior. You know, charging stations and heating, whatever it's called, electrification is very, very difficult. So, not surprising that the commission has started a lot of legislation on AI and just to recap

a bit, um last year there was the AI continent to build AI capacity at scale. And more than 200 billion, 200 billion euro will be mobilized for building AI factories and giga factories. The infrastructure for AI, I think the main infrastructure to develop large models, the compute, um and data and tools provide access to access to high-quality data. Second important legislation is once you have the infrastructure,

you also have to make sure that you create benefits for industry and users. So, it's the apply a strategy also launched last year, um moving towards AI from pilots to deployment, increased productivity in key sectors and support SMEs and startups gaining access to the the infrastructure and the AI data labs in order to really use and promote AI models. What's coming this year and that was mentioned

yesterday by my director Tibor, it's the European sovereign package with three elements, the cloud AI development act to scale up European data centers by a factor of three, to triple the capacity, ensure sustainability and security in AI deployment. Second one is and that back to energy, preparing a sovereign strategy for AI roadmap for energy is also coming up. Uh accelerate AI innovation, they moving the large models

for energy systems as well as small models to support in particular smart and intelligent demand supply flexibility and that's one of the key issues of decentralization. And the third element is you heard it yesterday, open different ecosystems to support open source for technology sovereignty and that's one of the key elements of my talk. uh to have an approach on that one, um particularly for industry you need

a structured approach. Um you mentioned the Eurostack already. Now, we have in order to ensure sovereignty and we have to cover the whole stack. The whole stack from the components of microprocessors, um the embedded software, um the edge computing, the networks and connectivity towards the the cloud, the applications and the application layer. a software stack to make it happen, you know, you need modular building blocks in

order to reuse what has been developed by others to have efficiency. And you have to connect systems, so that's the IoT part um and there you know, the complexity is increasing because you have to define the interactions and the interfaces of the different systems. And in terms of security, as you have you need some sort of governance models across different systems. You know, you mentioned the hourglass

model and then that has a a role played again over And last but not least, for software driven systems, you also have to make sure that you have the basic hardware for that and then we have a lot of projects supported under the Chips Act, the Chips JU. So, one of the benefits we expect from open source. First of all, it's the harnessing the power of innovation

because we see innovation now with the unprecedented speed. Um so, legacy systems have difficulties to cope up with that. And we have the advantage of the open source decentralized communities, the creators, the innovators to move faster, adopt technology, identify risk as well and um show implementations. So, um there we know we have a a lot of things where we can benefit from these communities. Second is inter-

interoperability of standards. Um we have recently seen that if you provide the code for an emerging standard, you know, you can really show it works or it doesn't work. So, it's much faster than formal standardization. And more important, you know, for the interaction of systems to test interoperability of standards. And that's, you know, is an added value for small companies and companies to use the standards because

if you only have one standard that in an isolation, you know, it has no value. If you test the interoperability and shows that it works through the whole stack, it has an advantage. Thirdly, efficiency through reuse of building components and there it's a bit um a plea for the the foundations to show, you know, the interoperability and that you don't have different isolated projects, that you have

a whole stack of projects which also uh collaborating. Third uh fourthly, um sustainability extended of product life cycles. You know all that, you know, if you you deploy hardware and then, you know, the change of regulation or the innovation, if you cannot use it anymore, I think open source has a potential, you know, can you still use extend the life cycle of hardware. That applies for PCs,

you know, putting Linux on that, you know, you know all that is more efficient, but it also has applies to charge stations, you know, in order to make sure they still are secure. So, US is demanding open source for charge station to avoid any backdoor in the system. And last but not least, you know, collaboration um has the advantage of skills and training for your people. Um

we have a parallel stream on software defined vehicle and I mentioned it because it's supported by the commission by A Cava, the European connected um and automated vehicle alliance uh looking at a whole software stack, the open source software stack for uh software defined vehicles across the software integration in the car from embedded systems to the infotainment and telemetry, the user experience as well as the hardware

and that's a a major adventure endeavor of the commission at the moment, but we have a whole stack. It's the SDV stack which is supported under Eclipse. So, that's a key example, you know, um where we have a project for full software stack led by by industry. And that is maybe an awesome example for other industry, you know, to have software defined autonomous systems built on open

standards and architectures. We have an open source stack so everybody can test it and that, you know, the different building blocks Uh and finally, it's also a leverage to integrate AI at the systems. I think without the software integration, you know, you cannot do any AI. So, um that's a key element and then, you know, you still don't sovereign computing and chips. So, we think that could

be a role model as a sovereign software stack also for other industry and that's a bit of discussion we have Eclipse, but also maybe a role model for other um foundations like Linux. Uh it's part of the sovereign package and we're looking, you know, also looking at applying that to other industry like software defined automation and software defined drones or software defined defense is uh in the

discussion at the moment or software defined charging as a stack for charging. So, and then back to to energy again, you know, what's happening there. Just to visualize a bit, you know, what is happening there. If you install a wall box, in the past, you know, it was a point-to-point connection, you get agreement by the energy system operator. Um this centralized operation is the one of the

past because, you know, systems are saturated, you don't get this connection agreement anymore unless you have a decentralized orchestration of the different charging and energy consumption. So, you have need load balancing between charging stations, you have to connect to residential batteries, you have to liaise in with big consumers like heat pumps. Uh and most importantly, you know, the simple thing is, you know, that you link the

charging to the solar, which requires, you know, again interoperability of standards. I think there's a connection, you know, also to the standard of the solar inverter. And to be practical again, you know, we have an open source project working with Linux, Linux Everest, on a a software stack for charging, which implement the ISO 15118-20, but also looks at the interoperability with the IEEE 2030.5 or the SunSpec

Modbus in order to link charging to solar, you know, the most easiest case. and then we, you know, they demonstrate the interoperability of standards, they give seminars and they really show, you know, where protocol is not ready, where you don't can do the mapping of standards. And it's great work, you know, of um getting the community together and look also, you know, not just coding, but also

providing the code for the standard with in order to facilitate facilitate the adoption of the standard and to have a um a print ready solution, deploy and ready solution. So, it's a reference hardware testing and a reference implementation what you have over there. Um and then you have a community, you know, you have easy access to either tailor-made solutions or, you know, to to do a mapping

to commercial solutions. Uh we get a lot of recognition over here um also from partly commercially um associations because also they want to see, you know, how to link to these standards. And if you have already a demonstration, it works a lot. And uh we will have another conference in November about the um EV leaders conference. Um no, V2G leaders conference in order to again, you know,

bring that to stage. And that basically is my last slide on the key takeaways, you know, what is our current concerns. Yes, I think we want to support the industrial transition towards agile programming interfaces, you know, the software is everywhere. Um but that's a baseline, you know, for AI based autonomous systems. For that, you promote open architectures uh across, you know, cloud edge infrastructures and down to

the chip level. Um the Eclipse SDV is a role model for really industrial engagement in showing, you know, with the to create an open source industrial stack. And uh and then as mentioned, open source is accelerating the standards adoption and interoperability of standards where give visibility to the lot of communities and industry and advice. And but and as we heard in the previous session they are challenged

to remaining about the governance across different projects, about security, trust and security, a trusted framework. You know who controls the trusted framework. We heard the issues of the manufacture and maybe we have some discussion after my talk now. Thank you so much for your We didn't do it on purpose, but the last line is hourglass model and I'm going to work talk about that. Okay. Because Ralph

was supposed to start first, but we made a mistake. Yeah, so the hourglass model we came up with the idea during the kickoff of CI sphere. It's a a colleague from the mirror who did this he he showed this and we thought maybe it's might a good tool to discuss together. Okay. I mean not only technical people but the the demand side and the supply side. Okay.

And we realized that actually it has been used quite a while. Okay. And there are many if you Google hourglass model you will see many. Now you'll see ours. Okay. But you can see that it was actually even a model model here. There's a mathematic formalization and they showed that this was the reason of the success of the internet using IP and also Unix. Okay. So um

the idea is we can do it to create an ecosystem which is synchronized. So where we have we bridge the the gap between the demand and the supply side. So here we have the managers. Here we have technical guys. Okay. But it has to work. So the consistency is ensured by the fact that you have standards. So that's what Ralph mentioned and also open source so available

solutions because if you have you create an ecosystem that is actually private you're not sure you you you could succeed but you're not sure you will succeed. So getting open source is a better guarantee to to reach people. Okay. So and I'm repeating some of the items which Ralph has mentioned about reusability and of course getting this consistency. So um examples. So we get into first the

standards and here you can see that if you look at this with this hourglass model where here we have infrastructure. Okay. So metal OS, embedded AI, energy OS. On top data analytics, use AI and things like that. So you can come up with standards that are useful and we're not reinventing the wheel. We are reusing things that have been created by people for instance minimum interval mechanisms

that come from smart city. We can reuse it. Okay. And below we have architecture one reuse. So architecture patterns is the thing we reuse. Okay. And then we have what we are doing in this project CEI cloud edge IoT assets. So we want to have reusable asset that we can use. Okay. And on top of it then you can have AI reasoning and things like that. And

we need semantic models. We need interoperability. So all those things we will address them and other things because here I didn't mention cybersecurity. Obviously cybersecurity is transversal. So the fact is actually in reality you have multiple ecosystems. If you try to put everything in one ecosystem which we'll have to do it's hard for the manager to understand. So you need to have pictures where you focus on

certain things for instance IoT cloud edge IoT data space and the trust framework. Okay. And we put here on purpose co-pilot. Sorry I didn't check with George but this is an example where actually the cloud edge IoT could leverage the things that co-pilot is doing. But of course obviously ideally we have to work together. So we compete in the sense that we provide different technologies and this

is the purpose of research but at the end of the day when you come to the market we should get together. Okay. So and we should have the same messages to the business stakeholders because at the end of the day they make the decision. So we need to work with them. Okay. That's super important and one thing that is also very important is if we keep this

hourglass model present the the business people, the stakeholders they understand what is at stake and what are the issues that want to solve. And it's also a reminder to the people working on research and technology that they have to care about the demand. Okay. Because if you don't do that then sometimes you do something and maybe you're out of the reality. Okay. So and you lose the

message that you want to have in common to create the things. because we do research we create a vision and the vision I would say it's a kind of technology push. Okay. You have to have this visible, explained and understood by the business stakeholders. But then you have to mature and some of us including the big companies they will go into a market push. Okay. So for

instance in in in France we had the market push about the the the smart meter. Okay. Because we had a first project in the year 2007 where we had like 80,000 meters. Okay. So this is again my market push and there was a decision and the decision is there and then and now it's deployed all over France the smart meter. Okay. So we need to go from

technology push, market push and market pull. And open source you can have here, here and here. So the open source is maturing at the same speed. Okay. So this is where we have success. Okay. that's why I have to have to be careful and I need the help from Eclipse Foundation and other foundations on that. So you first have the open source development maturity. Okay. So do

we do a good open source? But it could be a good open source that is only done on research. Okay. But then you have to which innovation stage you are. In research or industry. And you have to get from the the research to the industry exactly according to this line. And last but not least you have need to have community growth. Okay. So because the community has

to grow and it's a token it's an evidence that something is happening. Okay. So if we are not able to to make this community grow maybe you have an issue. Maybe not but we have to check. Okay. And this is the reason why we came up with a toolkit which we call the backbone toolkit. Sorry for the term. It's backbone toolkit for an open actually ecosystem. The

document is freely available for everyone and everyone can apply it. We are trying to apply it to learn about how we can have this hourglass model as a communication scheme between the developers and the demand side. Okay. So who uses toolkit? All of us. EU research project coordinator. That's not a joke. It's very true. Then policy makers, business stakeholders, project architect, product builders and things like that.

People. Okay. That's very important. Okay. And why use it? Strategic positioning. Understanding how we fit and how we can integrate smoothly for instance in the Euro stack vision. Okay. That's quite important. Okay. So and of course you do this because you can amplify the market take up because you are able to explain. Okay. So the manager can explain he used the hourglass model. the technical guy engineer

can also join. Okay. So things like that. And of course the idea is optimize the resource. That's why we have this narrow waist because this is where we focus on the thing we want to reuse and the projects wants to leverage. Okay. That's it. That's the goal. Okay. So we came up with four steps in order to do this because we cannot get the best hourglass model

to start even though when you start a new project people have in mind this. Okay. So we have a two lightweight phases where we were going to describe those hourglass models. So the first is identify the ecosystem. So stakeholders with common business objective. Okay. And this we know it from the start otherwise the project would have been accepted because you explain how you fit into the market.

Okay. So but it's lightweight because it's already there. Okay. So we just need to have it and show it as the objective. Okay. Then we go more in detail. We identify the the different ecosystems. You saw that we have several use cases, several proof of concepts. In OCI co-pilot probably there are several ecosystems. So we need them to work and to identify them and then we see

what's going on. Okay. And then projects is developing and we have the the part three which is actually CI ecosystem analysis and once we have done the implementation and the before we deploy and things like that we have a much better idea of what it is. Okay. And last phase which will be when we are mature about what we have implemented we have what we currently call

it trust label analysis but I could maybe just call it trust analysis because the label is the result of a certification. But in reality we want to do a trust analysis and want to identify what are the features that we need to have or we believe what we need to have to make it happen. Okay. And this is the kind of message we'll tell we'll agree actually.

We not tell we'll agree with the business stakeholders. If we say, "A, B, and C, maybe the market is growing." Okay? And this has to be a a common consensus between the business stakeholder, the demand side, and the supply side. Okay? And of course, supported by standards and open source. I think I'm done. Okay? On time, yeah. Perfect. So, we get to the next. Okay. >> [applause]