On Monday 11 September, IIT-CNR officially connected to SLICES-RI, the European research infrastructure which aims to guarantee the replicability of computer science research experiments.
On Monday, September 11th, the IIT-CNR officially connected to the European research infrastructure SLICES-RI, after a long study and testing phase. The technical operation necessary to connect the CNR computers to the infrastructure was conducted by the young IIT researcher Tathagata Maiti with the supervision of Andrea Passarella, head of the Ubiquitous Internet research unit, and his colleagues from the UdR Raffaele Bruno and Claudio Cicconetti.
To better understand the purpose of this 30-year European project, which aims to build a research infrastructure for European computer scientists studying networks, we asked a few questions to Andrea Passarella.
Let’s start by defining what a research infrastructure is and why it is important to a scientific discipline. “Virtually all scientific disciplines, such as physics, but also the humanities and environmental sciences, have their own research infrastructures, that is, a set of shared physical and intellectual infrastructures”, Andrea explains. “The most famous example of research infrastructure is CERN in Geneva. A high-energy or particle physicist with a particularly interesting hypothesis to test goes to CERN because that is where the international reference experimental facility is, where the tests can be done”.
The importance of these infrastructures, which allow synergies between different research institutions, is also recognized by the European Commission and different nations, which have put in place have specific programs to fund them. Information technology has always been present, but as a service until now: “Information technology is always present in research infrastructures, simply because you need networks to exchange data, systems to store it, to analyse it. But so far none of the existing research infrastructure was aimed at the scientific community of computer scientists and those who develop networks for research purposes”.
And that’s where SLICES comes in: “ESFRI is the European program for building research infrastructure, and it recently created a division called DIGIT, to fund infrastructure specifically for the computer science community.Computer science is a science as such, with its own scientific value going beyond that of service to other sciences. Since the last round of funding, two new infrastructures have been approved in this area: SoBigData and, indeed, SLICES”.
But what is it, in detail? “SLICES aims to create a distributed infrastructure based on a network of labs, in some cases building them, in others integrating and evolving those already existing, with the goal of researching future networks at the European level. There are many groups around Europe that have developed cutting-edge network technologies and systems to test them over the years. Researchers engaged in research, such as on Beyond 5G and 6G (by now 5G is something already achieved for us, no longer a research topic, but a commercialised technology), who may want to study a new algorithm, will be able to do so by taking advantage of SLICES. This way, using a known infrastructure, experiments are replicable, with transparent data, shared with the community, and based on state-of-the-art instrumentation and standardised methods”.
In fact, the replicability of experiments, one of the cornerstones of the scientific method, is one fundamental objective for SLICES. “In physics, it is normal that an experiment done by one researcher is then replicable by other research groups. In computer science it is particularly difficult to achieve it, especially in the world of networks, partly because of objective problems of complexity. If I have to do, for example, an experiment where I show that an algorithm on a wireless network works better than the existing one, it is very, very complicated to reproduce the same wireless environment conditions on which I did the experiment. It depends on how many people were there in the lab, when I did the experiment, the humidity in the air. There are so many factors and boundary conditions, which make reproducibility a problem”. Research infrastructures, in this respect, help a great deal in managing the conditions under which experiments take place, and thus their reproducibility.
SLICES, like all ESFRI projects, has a long-term perspective, in this case 30 years, which includes a long preliminary phase. “There’s a first phase of infrastructure design, which we have already finished. Right now, we are formally in the preparation phase, where we make sure we have the funds, the governance and management framework that are necessary to build the infrastructure, as well as the compliance with applicable legal procedures”. This step is necessary because, unlike other European programs, ESFRI does not directly fund its projects, which therefore have to find funds from elsewhere. In the case of SLICES, there are many ongoing projects financed by European research programs, like Horizon Europe, that support the infrastructure development. The Italian node is currently involved in SLICES DS (Design Study, recently expired), SC (Starting Community) and PP (Preparatory Phase), and two other, GreenDIGIT and SUNRISE-6G, that will start by the end of 2023.
“The bulk, however, must come from individual member states, who commit to funding the project, recognising its strategic importance”. In the Italian case, the groups involved are using PNRR funding to develop components that will converge into the SLICES program, exploiting significant synergies with the SoBigData research infrastructure (the other project approved within the ESFRI-DIGIT framework), as well as the Extended Partnership RESTART on the Future of Telecommunications.
In parallel with the preparation phase, the implementation phase will then also begin: “About halfway through the preparation implementation will start, lasting between 3 and 5 years, and dealing with actually creating the infrastructure. Only after this, comes the bulk of the project, the operation phase”. During the next two decades of operation, the infrastructure will be constantly updated.
Effective implementation comes through the development by each state of a national hub, which will then talk to other hubs. “We, as CNR, are coordinators of the national hub of SLICES. We are joined in this project by CNIT and CINI, the university consortia for telecommunications and computer science, respectively”. Some components of the national network are already in place, such as the 5G laboratory at the University of Genoa, participating as a member of CNIT. “We, at the IIT, are upgrading our labs, in order to establish the CNR site in Pisa, which will be completed in the next three years”.
One of the most complex things will be to harmonise the different laboratories already existing, embracing the technological standardisation defined by SLICES. The existing laboratories, compared to the new ones, have specificities that need to be taken into account: “The research environment is very heterogeneous, there are test labs with different levels of evolution, it is difficult to find a solution to integrate them suiting them all. You can’t just throw away an existing laboratory, such as the one in Genoa, to redo it from scratch with features more comfortable for all the other partners; you have to take into account individual investments, experiences, and expertise. Perhaps things done by one hub are not going to be 100% compatible with things done by, let’s say, French hubs, and that’s where the big effort of SLICES comes in to harmonise these differences as best as possible, maintaining a strong projection into the future, which will guarantee the full operation of the infrastructure in the next thirty years, while keeping pace with the continuous evolution of research in the Internet sector”.