Using the Hourglass Model for open source development in research

This talk discusses the application of the hourglass model for open source development research across various projects focusing on open cloud edge IoT, energy optimization, and agriculture. The speaker explains how the hourglass model can facilitate the creation of ecosystems that bridge demand and supply in technology, emphasizing the need for interoperability and reusability. The presentation also highlights specific projects like OCI, aimed at cross-domain data sharing, and AIOT, which focuses on digitalization in energy systems through IoT solutions. Additionally, the EQUIS project targets cybersecurity with a focus on interoperability and secure data spaces. Throughout the talk, the importance of open source solutions and collaboration with organizations such as the Eclipse Foundation is underscored.

[music] >> So, title of the presentation is using the hourglass model for open source development research. You heard this morning uh Tibo Klana said open money leads to open source. So, I would say open money leads to open Then uh we get to standards and then open source. Okay. So, uh this is our outlook. We're going to show to you what is the hourglass model and see

how it is being used in four projects. Hourglass model was invented a long time ago and uh there was even a paper published in Communications ACM 2019 to show how uh you can use it to understand how you create the ecosystem. How the ecosystem will grow, okay? And two examples we'll use. One is IP and the other one was Unix, okay? So, uh we applied it. And

uh the purpose is now that we want to create an ecosystem where we bridge the demand and the supply side. Exactly the previous talk we're talking about research, but we need also the industry. And this is this bridging that is most important, And we need to find the drivers for usability so that actually uh the demand will use reusable technology. In order to reach that, we need

to create a consistent way of uh creating the system from the stakeholders to the capabilities including the standards and then open So, uh first of all, we need to understand that there are layers because on top you might have uh users and below the infrastructure, okay? So, you want to layer the key enablers. Then uh because you want to put the things together, you want to foster

interoperability and reusability. Then uh you want to leverage regulation, harmonized standards. This is fresh, really new. So, we need to do it uh now, okay? And uh we want to leverage standards, okay? So, and the last thing is when you have this, you want to make the thing happen, so you use open source. So, here's an example of reusability that was created in the CEIS FEAR project

where we focus on CEI, cloud edge IoT. You see on the left, you have stakeholders. So, that's why you have only uh business people here. On the right, you have capabilities, technology, starting below from Internet of Things, cloud edge IoT, uh digital twin platforms, data analytics, user interfaces. Um here's an example of standards uh where uh we can focus on CEI assets, architecture patterns, minimum interoperability mechanisms,

and on top, semantic models for the users. So, you can also uh use it for regulation. Below, you have uh On top, you have regulations, and you want to have reusable conformity mechanisms. But, there are multiple ecosystems. So, uh we have to bet about where we we we create the growing ecosystem. So, here is an example, which I believe is where we are currently navigating with the

projects we will show later. Uh we have below IoT domain ecosystem, the cloud edge IoT ecosystem where we bridge the gap between IoT and the cloud. You have then the data space ecosystem when you want share data, and you want to have something that is trustworthy, so you have the trustworthy framework But one thing that is important is we do research. So when we start, we have

technology push. So the ecosystem is a technology push Then you mature and you get to a market push because you want to push the market, okay? And then if you made the right bet, uh you will have a market pull. So the market will will take it, okay? And the whole problem for us is to get to the last stage, okay? And we need to find all

the tools including open source to make sure that people can say, "Okay, now I can use it." Okay? So uh when it comes to open source, you have different things to check in this maturity adventure, okay? So first of all, is the open source itself mature? Okay? That's why we need Eclipse, the foundation to tell us, "Yeah, here is good, but here you should improve and things

like that." Okay? So for instance, IPR issues. You need to to to solve them, okay? Then you also have the innovation stage. So the the industry the research solution could be good, but not yet mature for industry. And the other way around, okay? So we need we need to check those things. And last but not least is a community growth. So if you have a system that

is good but only used by three people, it's not the same. So you need to make sure that the community is growing. So I'm going to give the floor to uh Cecile for one project which is OCI. Thank you very much Antonio. I hope you can hear me well. So now I will present the application of the hourglass model methodology that Antonio just presented to the OCI

project, uh which is a European project coordinated by UPV Universitat Politecnica of Valencia. So OCI stands for stands for Open Cloud Edge IoT platform uptake in large-scale cross-domain pilots. So, the mission of this project is to drive the rapid adoption and scaling of cutting-edge solutions for energy optimization, thus enhancing Europe's competitiveness and enabling strategic autonomy. So, uh as you can see by this list of some of

the main OCI innovative outcomes, this project has a strong focus on data sharing and this cross-domain data exchanges. And that that's is also illustrated by the the goal to develop a cross-domain continuum ontology. And this project will offer a federated marketplace to allow users from uh all over Europe to discover, select, and allow installation of solutions developed in the project either in the large-scale cross-domain pilots or

in the open call projects that are taking place right now. Uh the solutions from the OCI project will be applicable in six key strategic verticals, as you can see here, electricity grid, electromobility, software-defined vehicles, as we've heard a lot about this already in this event cuz it's very a growing topic, agrifood and agriculture, logistics and urban environment. As I mentioned, the project is coordinated by UPV and

started in January 2025. Uh here you can see some information about the eight large-scale pilots. So, so that they are located all over Europe. They all have their main topics all related to energy optimization, flexibility. Some of them with specific work on vehicle to grid, federated learning, and a lot of different topics. But what is very important also to explain is that there is a strong focus

in this project on cross-domain data exchange as you can see by the relationship between the different pilots. So they will exchange data but also cooperate on AI models and also share best practices, user engagement methodologies, strategies, and everything will be cross-leveraged around the project. So how does the hourglass model application in this pilot in this project look like? the So the OCI project started to adopt this

methodology to describe the CEI ecosystem and particularly the ecosystems of its large-scale pilots. This allowed to identify the minimum interoperability mechanism that is the most needed in this project and this is, as illustrated in the middle, finding a way to give access to CEI assets to both the users, so the top part of the stakeholders as explained by Antonio, and also to the infrastructure providers. This way

they can all implement and make available the capabilities that they are responsible for. This will be achieved through a CEI asset marketplace that is developed by the project and I will result in a the platform that can be used for instance by energy management system providers in order to manage a energy in the different different pilots. And this analysis, which is still a work in progress, also

allows to continuously identify the open source needs of the project partners. In this case we do it in a collaboration with Eclipse Foundation, which is also a who is also a project participants. And one of the main target that has been identified is the development of a CI asset builder, which will indeed be open source. And it also needs to be compatible with Eclipse data space components

and also the TM Forum product specification and uh and you preview will explain more in a second talk tomorrow about also the open source targets in the different pilots. Thank you very much for your attention. I'll introduce your colleague now. Good afternoon everyone. I will present the AIOT project and all this consortium. So, AIOT means holistic approach toward empowerment of the digitalization of the energy ecosystem through

adoption of IoT solutions. It's a project that was started in January 2024 and that aims to deploy IoT assets at different level of the energy system from behind the meter to the TSO level. It also aim to add intelligence at the and cloud layer and to bridge the cloud edge continuum. The project turn around smart energy, smart grid, semantic interoperability, grid management and optimization, flexibility, AI energy

communities. For this, the main component are data space connector, edge cloud services, app store, computational orchestrator, semantic enablers and AI ML services. Of course, we have pilots. We have six pilots spread all around Europe. Two are more focused on grid management and optimization, two on energy flexibility energy communities and two of them are a bit hybrid between those two topics. On this slide you can see the

concept concept to the AIOT project. On the left side you have the stakeholders. I want to highlight the key players, the key stakeholders, the grid operators, TSO and DSO, as well as the energy community users and operators and the flexibility market users and operators. The key aspects we want to highlight with AIOT is the AI based service library for energy data space. That's the key point of

the project. And to allow this, we we are going the implementation of capabilities as like an app store for all the services to make them available throughout the architecture for all the different stakeholders. Also some cloud and edge AI based services for energy grid optimization, grid infrastructure, flexibility and energy communities. And data space connector is is used. It's based on Eclipse data space and also the data

space protocol, the DSP. Semantic interoperability layer is in place and that's one of the key elements On the right side I want to to conclude by the highlight by an highlight for the different proposed open source aspect of The first one is a library of semantic interoperability enablers. The second one is a library of AI based edge cloud services. And in the project, we are IDS compliant.

So, we we are data space compliant following IDS uh standards and framework. And uh using the last version of the data space protocol. We can also just mention briefly all the items like an open app store, uh an open orchestrator, and open IoT solutions. Of course, the list is uh long, but those are the highlights. I can let the floor to my colleague for the EQUIS project.

Thank you. Um okay. So, I'm going to present the EQUIS project, which is uh let's say cyber security project. Um it is a project that is funded by, once again, the European Union, and it covers uh three different uh type of tools. The first one is the Uwallet We also have the development of secure European data spaces. And one of the biggest purpose of this uh project

is to fight uh cyber threats and identity-related crimes. Therefore, uh once these uh different tools are developed, the idea is that we we can uh apply them in the cyber threat uh AI-based uh pilot, and also in a pilot that is uh um focused on health, more specifically, but the the use case in this health pilot is uh lungs uh diseases. So, if you want to read

a little bit more about the project, there is the the link of the project is not that different. It is stands for the the the sweets, of course, but uh it is the Philly based for sovereign identity and privacy for a tool set. Uh we have 12 partners. Uh now we can say 13 cuz we have one that counts for two. Uh they're all distributed in eight

different member states. Uh the focus once again is interoperability, scalability, uh usability. Um one of the purpose is of course to have open source adoption, which I'm going to explain with my hourglass model here. Uh in the case of Eclipse project, the idea is that we're going to as all of uh projects where we apply the model, we have to identify the stakeholders and the capabilities. The

stakeholders in this case are the ones that can offer us trustworthiness. So, for instance, we have uh the applications that uses identification as the source to to allow the the users uh to the system. Then in the let's say the the narrow layers there where we we we tend to say that is the minimum interoperability mechanisms as is the layer that will allow us to have a

common communication standardized way of of of talking. Uh in that one, we will think of all of the uh security and privacy compliance uh stakeholders that can somehow help us in this standardization section. And then uh in the infrastructure like close to the infrastructure side, we will have all the data and the actual Um here is more about the who's going to provide a wallet, who's going

to provide the cloud services. Um in our case, uh the pilots are more uh uh on site placed since the pets we're using are by instance biometrics, uh federal learning. And then that lead us to the capabilities. So, um as I was mentioning, the we have the wallets. And in our case, one of the capabilities will be that for the e-wallet, we align with the identity frameworks.

Then for the interoperability layer, we can think of once again the self-sovereign identity and the privacy-enhancing technologies. Uh what for instance we can think of the ISO IC 27, the OpenID, and all of the ongoing already um standards. And also all of the different acts that can help us like the CRA. And then we have the infrastructure management. In here uh is more about what will uh

help us managing uh the physical infrastructure. So, we have the PKI, the ELTI. And uh that could lead us also to data On the open source side, uh we can think of the system helping us in the interoperability sense of having building blocks that uh completely uh compliant with the data space with the data spaces um already community working on open source communities. For instance, uh one

of the one of the partners of the project is IDSA who is working in the least uh data space components group. And also we have like a strong links that we can say allowed us to contribute in the least foundation data space initiative. Uh just I want to add uh there are five sister projects and uh we will enroll them. And the goal is to provide the

future library of pets for data space, okay? So, this is the goal and we have a bet with with Azahara that if we have this, she pays me a bottle of champagne. Uh be careful. Uh hello, I'm Estibaliz and um Yeah, now we are changing completely uh sector area. We've been talking about projects on energy, e-mobility, privacy and data protection, trustworthiness, and now uh we come to

arrive we arrive to agriculture. Okay, so technology is paid project which you already have there the roll up twice. I've seen it. >> [laughter] >> So, do not miss any information, okay, >> so here's okay, technology for agriculture in this case. What technology? Drones, data spaces, digital twin. And how? Through open source strategy and use. Uh SPADE is a four-year project Horizon Europe uh it's uh has

21 It will finish this year at the end of August. Maybe we'll listen a little bit uh depending on the European calls that we have had for applications. The main uh goal of the project is technical speaking is the implementation uh ecosystem, a multi-purpose physical cyber agri-forest drones ecosystem. Okay, this is for the governance and also for environmental observation. global objectives, of course, of the commission, we

have the enhanced ecosystem farm to fork. But also to develop the platform uh which is multi-purpose generic uh to manage the drones and also use the data for uh on top of that build a different of course, we provide open source Uh also Eclipse Foundation is uh a partner of the project. So, they can check as well. it has uh three demonstrations pilots in livestock, agriculture, and

forestry. Uh we have two open calls. Uh now it's the second open call. They are deployed into different applications. And also there are some synergies with the two other sister projects which are Chameleon and Icarus. Most of all on the crop monitoring to detect diseases, to track the health of a plant and also for tracking different animals, okay, livestock and of course which is we can see

the start of it. The drone spraying agriculture is something that every cooperative or every other farmer wants to to use. I know because I come from the rural zone so I know. And here we have just in a brief outline the three pilots of this beta. The first one the forestry in Norway. Swan for inventory there are different drones created and built. Larger for drone leader, a

scanner flying, for the front or lateral for also lifting the trees and planting, okay, the planting drone with machinery solutions. Then we have the agriculture one in Majorca, Spain in two places. One in Soller and another in Sa Pobla so we have different types of plants because in the first one we have olive and citrus fruits trees. The other one is potatoes. And this is to use

the drones for crop diseases detect, soil characterization. To support the decision making. To the end users. It could be the farmers, the agricultures and the cooperatives. Which treatment to do and to save efforts, time and money. And of course this is the high precision agriculture we can say there for fumigation and irrigation just in the drones. And the third one is in Greece, livestock. This is for

the optimization of animal grazing and health care management with drones. So it can well, tracking and real time the animal with IoT chip collars. Uh but also is detection of maybe I would say abnormal uh or specific behavior which crazy moving circles limping so they can detect that the animal uh dead, badly injured, whatever. Okay, so that's the final goal. And just having a glimpse here of

the how we apply the wireless model in this case in this project. At the left we have the stakeholders. we have maybe it is ready to start with with the middle that you can see in bold. Uh in the middle we have the minimum interval interoperability mechanism which is the access to agriculture data using the space digital twin. So, we have the stakeholders, the open source digital

twin platform provider. And also the capabilities of the digital twin data application which is important. On top of the open source digital twin platform providers we have the end users, ABB developers, data space and platform providers. And if we go to the bottom we have the infrastructure infrastructure providers, the ones that build design the drones, IoT devices depending on the application. Centers, etc. If we go to

side we have capabilities. We have already said one important one is the digital twin data aggregation. Data space analytics and machine learning algorithms to analyze the agriculture data for specific making decisions. Uh different applications on that. The edge computing drones operation and swarm flight. And of course IoT based sensing and actuating. And we will finish with uh okay, open source. We just uh the core of of

this presentation also together with the hourglass model Here uh a Speed project is already an Eclipse project, if I'm not wrong. Uh it's just finishing the last things. It is also uh already available. We have also different folders with the uh different open calls winners And one thing very important is that uh we're implementing uh BASICS which has been just already before only implemented and used in

a smart manufacturing. So now we're changing using the BASICS open source in another field sector completely different, in agriculture. I don't know if you So this is the whole picture of the project very uh brief. And uh I will let Antonio to finish with CI toolkit explanation. So thank you. So uh this is an exercise for all of us uh to uh bridge the gap between the

demand side and the supply side. We use the hourglass model because the focus on what is the jam that we are producing at the end of the project, okay? So uh the CI sphere came up with a toolkit, a methodology allowing us to think about it, okay? So uh it addresses uh ecosystem fragmentation. It allows us to understand what it is worth to reuse. It is concerning

all the stakeholders. We start with the EU research project coordinator, okay? Uh he has a good idea, but then we have the other people, the policy makers, and and and so forth, okay? So we want to have uh it to allow us to have a strategic positioning of the ecosystem, of uh what the project is providing, and uh what it is going want in terms of open

source, okay? So that uh we reuse and we have an impact, okay? So and we want, of course, to amplify the market take-up coming from the technology push, market push to market pull, okay? And of course, we want to optimize resources. So this is the uh toolkit. You have here the uh URL where you have the document the the document which is explaining it. In a nutshell,

it has four steps. So we call it toolkit one, toolkit two, toolkit three, toolkit four, okay? Uh the first two, toolkit one, toolkit two, are lightweight. We do it them quickly. We have like a session discussion, we brainstorm, and we have the result, okay? So first of all, we have to identify the ecosystem. So in this ecosystem, first of all, we identify the stakeholders with their objectives.

This is the most important, okay? Then we go a step further where we have a high-level ecosystem discussion process including open source. So we're going to get further, what are the layers, the stakeholders, the concerns, capabilities, standards, and open As a first stage because we are starting in the project, okay? Then we have to mature and you have toolkit three and toolkit four, which are following the

development of project. They take time because we want to validate, we want to make sure that our first idea is good. So we will mature. So uh we have uh toolkit three, which is in-depth ecosystem analysis including the ecosystem, okay? We do this while we are developing because we will realize that we made mistake, we want to switch, we want to change things, okay? So that at

the end or nearly the end of the project, uh I would say before deployment, okay? So before the the the proof of concept, then we have very good idea of what it is we are doing, okay? And this is uh toolkit three. And then we get to the last stage, which is very important, okay? Which is how do we create an impact and we make sure that

identify the goodies, the right features that are making up our ecosystem and we would like to to verify, okay? So, if we go to certification, we would do this, okay? And we call this actually trust label analysis because we are identifying the good things in our implementation that will highlight and tell the business manager that because of them we'll have a good ecosystem, okay? So, the TK4

is very important and CIS fair has this ambition to create the ecosystem, the sorry, the trust label for CEI. The CEI asset will be associated with this trust label. But, we don't want it to be just a marketing label, it has to be technical. It tells us that we have reached the right quality to go to the market. So, yep. We're done. Questions? Questions? One questions? No?

Not acceptable, the second one? Uh thank you for introducing this wonderful concept to work on the bridge of technology transfer from research into industry. Um one question I would have from the 4K tool 2K from the first tool 2K you presented and especially the earlier version where you spoke about digital identities is is if at any point you're doing some kind of I would say LC impact

assessment uh especially given that you highlighted using SSI technology with this now known to actually need to issues of national security by the data broker files and also systemic discrimination of various kinds of users. There is an example out of Hamburg in Germany where users of public transportation get a worse credit score because they have installed the public transportation app and are trackable by unique identifiers. And

um I would love to hear how those So elsic criteria flow into the standards you're defining because the populace at large is usually not represented by a stakeholder. First of all, the answer is we should do it. And even more than that, we shall do it. Okay? This elsic stuff, okay? And so in the way it is presented, we are presenting the concerns of the stakeholders. And

you can see that sorry, if you get to let's say here you have to have the elsic people, ethics, and things here. And the policy makers. And then it can flow, okay? Because then you can the requirements that uh the stakeholders are having on the system. Yeah. Maybe if I had like sorry, like specifically on the on the case of the self-sovereign identity which I think is

one you were thinking of. >> Uh well, here um Yes, of course. We need to tackle that and I think we all here like hearing about all different conferences are we concerned about it. But they just to separate it a little bit, the hourglass model allowed us to identify what in the whole ecosystem we should target first. So here we we're thinking about okay, we we're going

to talk about we have all of those issues, but who's going to who who's the concerns? The concerns are for the SSI providers and the people who are integrated the the the bits. It starts with the fact that the OpenID Foundation is not a standards body and you list them as Because in the in the in the narrowly layer, they they identify who can help us in

the in this platform. >> problem I'm trying to address. The OpenID Foundation is but let's maybe make the society's problem. >> that's you are right. The OpenID is actually has a liaison with 27 that we mentioned it. But we here we have capabilities and in a capability sometimes use OpenID solution. But yes, it's not a standard per se. It could be a something that is by practice.

The point is that so to to make this transparent, I was a member of the European Union ad hoc technical advisory group. Yeah. And we became aware of the fact that the OpenID Foundation seems to be running a coordinated effort by its industry partners to actually make their standard their concept which are used to label machines to label humans. And um my accent might have give it

away. I grew up in Germany. Don't like the idea of numbering people. Okay. So obviously those issues we have to take them into account when we do the analysis. Yeah. I would fully support what you just said. Okay. So it turns out that maybe in the project we have here we didn't address that. So yeah, that's the one of the point that is important. It could prevent

the ecosystem to grow if we we just forget one of the things. So I just fully agree with you that when you have societal concerns at the stakeholder level we need to take that into account. Yeah. Um if we are late, I think I can take