Dr. Kei Karasawa joined NTT Research at its inception and serves as its vice president of strategy. From 2015–2019, he worked with the R&D planning department at NTT, where he built cooperative relationships with NTT operating companies around the world to deploy NTT R&D technology to global markets. He led applied R&D at NTT EAST from 2011–2015. Prior to that, he researched network software technologies, implemented patented software, such as security and distributed systems, and developed commercial services for the Next Generation Network. In 2005, he conducted basic research on cryptography and information processing as a visiting scholar with Professor Dan Boneh in the Security Laboratory at Stanford University. He holds a Ph.D. (engineering) in data-driven parallel computing technology and has extensive knowledge and experience in information processing-related areas, from basic technology to applications. We spoke to Dr. Karasawa about his background, role at NTT Research and thoughts about bringing research to market.
Could you tell us a little about your doctorate and how that research might relate to your subsequent work in R&D planning and development, and now strategy?
The research theme of the laboratory I belonged to in my doctoral studies switched from parallel computing to communication systems when there was a change of professors. So, it was difficult to continue my original research on parallel computing, but the new professor agreed to supervise my research under the rubric of signal processing, so my Ph.D. became more interdisciplinary. Through this process, I learned that even though goals of research may differ, basic technologies often have something in common, and that new research elements are easy to find in these overlapping or boundary areas. After joining NTT, I researched and developed IoT services for the IPv6 Internet, applying the basic technologies of parallel computing and signal processing to processing within devices and routers. Based on these experiences, I believe that more basic research in quantum physics and information mathematics can be adapted for use in a variety of ways as we move forward, toward a ‘beyond digital’ era.
In a typical technology company, the VP of strategy would take the long-range view. Is that different at NTT Research, where the scientists and lab directors are already doing basic research on a long term-horizon? Is one of your jobs to help find strategic ways to implement technologies in the nearer-term?
When considering technology development strategies in the near-term, say within three years, you tend to use a macroscopic lens to select technologies that can be implemented in society, based on current trends and the scale of investment in each area of research and development. But there are signs that technology will bring about discontinuous changes in the long term, perhaps ten years into the future. These signs are present in the cutting-edge research at universities and other institutions. To be more strategic, it is therefore also necessary to have a microscopic perspective on this kind of basic research. Fortunately, NTT Research and NTT R&D Japan have a large number of professional researchers who are on these frontlines. By understanding their research situation, we can think more concretely about the possibility of future technology. If there are useful byproducts from these cutting-edge research activities, we will be able to present customers with a theoretical and feasible technology roadmap using both macroscopic and microscopic perspectives.
As a visiting scholar, you did some work on cryptography with Professor Dan Boneh at Stanford in 2005. That was a few years after he came up with his implementation of identity-based encryption (IBE), and about the same time that Amit Sahai and Brent Waters were proposing their version of IBE, which became known as attribute-based encryption (ABE) – and last year won a Test of Time award. How does that work with Professor Boneh relate to your interest today in trying to find ways to deploy ABE?
When I was at Dan Boneh’s Security Lab, IBE was already in the implementation phase, through Voltage Security Inc., which Professor Boneh had co-founded. However, cryptographies based on the Pairing function, like IBE, were intensely discussed in other contexts, such as computational methods for encrypted data using homomorphic cryptography and functional cryptography. Actually, I couldn’t keep up with the cutting-edge discussion by Professor Boneh and researchers in the lab, as he converted equations from one end of three white boards to the other, but I could see the enthusiasm of the lab members. And I still remember one of my fellow Japanese researchers saying that Brent’s ideas were great. Cryptography using the Pairing function is considered to be the core of future security technologies. I also believe that ABE, using the foundational Pairing function, is one of the byproducts of this research that will be useful for future security platforms.
Why is now (after 15 years) the right time for ABE? What kind of drivers do you see for its application and deployment?
I believe that progress in data utilization and changes in the New Normal beyond the pandemic will be key factors. With the advancements of AI and IoT, there is a growing trend to consider data as an asset. In addition, in the ‘remote society,’ which is considered to be part of the New Normal, many activities are being brought online. So, various activities can now be collected as data. Here’s one idea: In the field of higher education, online lectures by professors and student responses (and perhaps even test scores) could be analyzed to create insights into performance and educational effectiveness. But this is sensitive information. As data becomes more diverse, privacy protection laws are being enacted around the world, and the need to protect data created by people and objects is becoming apparent. ABE is a cryptographic method that can embed access control into the data. In the past, it was sufficient to implement access control in storage management systems, but in order to utilize one part of various information or data sets, such as activity logs, while protecting privacy, I believe that ABE technology that can embed access control in the data itself will be required more often.
In your video discussion, you indicate that ABE is well-suited for smaller businesses. Why is that the case? Is that a template for applying other NTT Research-related technologies?
ABE is an elegant implementation of access control based on advanced mathematics. If you’re a big company, you can spend a lot of money on an existing system and add features to it, and on the surface, you get the same results. I believe it is important to establish basic services in new niche business areas where ABE is needed for its privacy protection and other aspects. The ABE implementation is simple, so we can start small. But if deployed in a large-enough ecosystem, it can then be scaled out and spread widely. Although there are no templates, I am discussing specific deployment strategies in collaboration with NTT Laboratories, which is developing security technologies in Japan.
Is there anything else at NTT Research that is relatively close to implementation?
In addition to the Cryptography & Information Security (CIS) Lab, the Physics & Informatics (PHI) Lab and the Medical & Health Informatics (MEI) Lab are also conducting cutting-edge research, and I am trying to identify potential by-products from them. Some candidates have already emerged.
Are there any milestones you would like to see NTT Research reach in 2021?
First, focusing on ABE, I will conduct demonstrations, PoCs and other activities with the aim of enabling customers to use our technologies jointly with our operating companies. I will also start standardization activities, which are necessary for technology deployment in many industries.