Director of ISP RAS - on the immutability of the pre-sanctions strategy and "thin spots" in Russian technologies
Partners of the Institute of System Programming of the Russian Academy of Sciences Samsung, Huawei, Bentley Systems have confirmed their plans for long-term cooperation, said in an interview with "Gazeta.Ru" Director of ISP RAS, academician Harutyun Avetisyan. He also shared his opinion on which domestic processors should be supported, what exactly will be difficult to import from the software and why a Trusted Artificial Intelligence Center organized on the basis of the ISP RAS is needed.
- Has the life of the Institute of System Programming of the Russian Academy of Sciences changed with the introduction of sanctions?
- Recently, young guys came to my office and said: let's write in the next collection of technologies from above that our strategy is unchanged. With the introduction of sanctions, they assessed that everything we did was right. So there is no need to radically change anything yet. It's inspiring.
- Did the contracts of the ISP "fly off"?
- If we talk about the financial component, so far
The budgets of this and next year are fully aligned, there are firm agreements and there is some margin of safety. Moreover, in February-March we signed several new agreements on our work. In particular, full-fledged cooperation with foreign partners continues.
- Every day we see that several more technology companies have left Russia. What, from your point of view, is good for Russia in this process, and what is bad?
- Breaking up any relationship is bad. But if we talk about our personal experience, we see the following trend. Here we had a very good customer - the Canadian company Klocwork, which was bought by the American company Rogue wave. After the events of 2014, they gradually froze, and in 2018 they finally broke off relations with us. Is it bad? Badly. But the team that did these works was completely retained at the institute, which gave us the opportunity to start a new research direction. These employees are currently working on the application of artificial intelligence methods in software engineering.
- What is it?
- This is, let's say, the science of how to use an engineering approach to write complex programs so that they are efficient and reliable. The use of artificial intelligence in this area is a new direction, few people are engaged in it in our country. For example, how to automate and optimize testing of user interfaces using artificial intelligence - in other words, how to replace a human tester with software. There are almost no industrial solutions to this problem. We have proposed an approach to how to do this.
Or there is another direction - the use of artificial intelligence in tools for developers. For example, the creation of specialized tools for code analysis and the search for specific errors in software. These are the issues that the newly released super-qualified people are dealing with at our institute.
The breakdown of relations is a crisis situation, but any crisis is not only great risks, but also great opportunities.
- From your point of view, what are the "thin spots" in our Russian technologies? What do we need to import right now?
- First of all, I want to emphasize that there is no task to completely replace everything with imports. No country in the world at the current stage of technology development will be able to do this. The task is to ensure Russia's technological independence. If we talk about "thin spots", then, of course, these are factories for the production of microelectronics, microcircuits. Another critical "thin spot" is CAD.
- What is it and what are they for?
- These are complex computer-aided design systems that allow you to create any electronic systems, for example, processors. Now, even when designing the domestic processor "Elbrus", foreign CAD systems are used.
One example is BIM technologies (Building Information Modeling). These are technologies for modeling processes and products in architecture, construction, infrastructure engineering and facility management. That is, when everything around is first developed digitally, so that it can be understood how it will behave in certain spatio-temporal conditions.
- Is it necessary?
- Without such systems, long-term competitive development of entire sectors of the economy is impossible. For example, construction. Otherwise, it may turn out that they built a theater, but they can't bring the piano in, because there are no such passages. Or something was poured with concrete, but it turned out that they forgot to make boxes for wiring, and then they need to change it all. Now in the field of BIM technologies, both augmented reality and drones that control the construction of the facility are already being used. It's such a big world. But in principle, we have some developments in our country, and if such a task is set, it will be possible to ensure technological independence in a few years, and in a few more years - to compete at the global level.
- Well, everything is fine with cybersecurity in our country, right?
- Now the trendsetter in the field of cybersecurity is the United States. Both Europe and China use American technologies to ensure the life cycle of safe software development - from design to operation.
For example, our static analysis tool is implemented in Samsung, and here we are quite competitive at the global level.
- From what you have listed, the most labor- and financial-intensive is the establishment of microelectronics production?
- yes. And here, as I believe, it is not worth chasing miniaturization.
It is necessary to create such technologies as soon as possible (for example, for 90 nm), because this will cover most of our needs, and we will still be able to bring the rest through any sanctions.
- And which processors do you consider the most promising? "Elbrus"? Baikal"? RISC-V?
- I think we need to invest in all three developments.
- RISC-V is an open and free instruction system and processor architecture. It is created by the whole world. Don't we risk investing and then being disconnected from it as a result?
- It is unclear exactly how we can block access. Since this is an open solution, it means that access will be closed to everyone. However, the "central headquarters" may stop accepting improvements and changes to the code from us… This, of course, can be a problem.
RISC-V, among other things, from my point of view, is also a broad educational circuit. The openness of the architecture makes it possible to implement research and educational projects. You can change the architecture of the processor, get feedback, involve dozens of design centers in its creation... It's all by analogy with open source software. Many scientists, students, and specialists participate in the creation of the processor, conduct research and thus improve it. Without this, long-term development is impossible.
- Why can't this be done within the framework of the development of "Elbrus" or "Baikal"?
- Because they have a closed architecture. But even if they open it, you need to attract people there so that they start using this platform. And RISC-V has a fairly large ecosystem: serial processors, microcontrollers, including neural accelerators, are produced on the basis of their architecture.
- Currently, various domestic operating systems are being developed. Which ones do you think are the most promising?
- They are all of a different class. If we talk about real-time operating systems, several organizations make them − these are NIISI RAS, Sukhoi Design Bureau, RKS, GosNIIAS, etc. Our institute participates in most of these works, in some our JetOS (an operating system for civil aviation avionics developed at ISP RAS - ed.) is simply taken as a basis. With this segment, more or less everything is obvious, and we need to develop further.
But we have many companies there that are actively involved in this competition. The Astra Linux solution is known, the other is from Basalt, and much more. The third separate segment is everything related to mobile OS. Our most famous system is this is the "Aurora", which is used in phones and tablets. Separately, we can say about Kaspersky OS, which can be either a full-fledged operating system or a substrate for another in order to provide additional security. Even what I have just listed shows that
- A Technological center for Linux Kernel Security Research has been created on the basis of your institute. What is it for?
- The main task of the Center is to increase the security level of Russian Linux systems. Historically, we have worked a lot with the Linux Foundation [Linux OS Development Consortium] on vulnerability detection, so we have extensive experience in this area. And we have a full range of our own secure development technologies.
To be more specific, then
In addition, it is very important that a community of domestic companies is being created around the Center. Among them are Kaspersky Lab, Security Code, Astra Linux Group, IVK, RFNC-VNIIEF and many others. This community should ensure the maximum level of trust in the core. Together we plan to develop patches to eliminate vulnerabilities in the Linux kernel, to fill the threat database of the FSTEC of Russia. That is, to work in a triangle model: a regulator (the state), plus us as a non-profit research organization, plus a business that provides feedback for competitive development.
- ISP RAS is a part of NCMU "Digital Biodesign and Personalized Healthcare". This world-class Center was created as part of the national project "Science and Universities". What kind of projects do you take part in?
- Systematic data analysis in medicine requires not only modern approaches in the field of IT infrastructure and software, but also close interaction with medical experts. As part of the creation of NCMU, we are working in a consortium with Sechenov University and other organizations; we are making a platform for aggregating large medical data based on our Asperitas cloud platform. Besides,
A web-laboratory for ECG analysis has been created. Digital medicine is a promising direction, and I am sure that there will be even more projects in the future.
- Your project of the Trusted AI Center was recognized as one of the six best according to the results of the application competition, which was considered by the expert council of the Analytical Center under the Government of Russia. What are you planning to do?
- Our main goal is to create methods and appropriate tools for the development and verification of artificial intelligence technologies with the necessary level of trust. Now artificial intelligence surrounds us everywhere: these are facial recognition systems, autonomous driving systems, etc. But if everything is more or less good in the field of traditional software development (without AI) - research is underway, there are code analysis tools, there are techniques, then in the field of AI there are a lot of tasks that need to be solved. Technologies are needed to counter specific threats - for example, various attacks on machine learning models. Highly qualified specialists are needed to develop such technologies. There are many tasks, and we are going to solve them. And not alone - among the partners of our Center are MIPT, MOC MSU, MSU Mehmat, Nizhny Novgorod State University, as well as representatives of the industry, for example, Kaspersky Lab, EU-Leasing and others.
Anna Urmantseva