Regarding current network technology, many researchers focus on how to improve network efficiency. By contrast, at Keio University, the Kaneko Laboratory, in the Department of Information and Computer Science (Faculty of Science and Technology), is doing research to construct next-generation network environments from the viewpoint of applications.

“If we think about the evolution of networks up to now, first, there was telegraph technology, then telephone technology, then the Internet, and now, we’re researching what comes next. But if you ask what the innovation was when networks developed from telegraph to telephony, it was a new way of using networks: The biggest innovation was being able to communicate spoken words to the other person. When networks evolved from telephony to the Internet, the innovation was being able to send data directly from computer to computer. That’s an innovation. Without looking at the question of what’s new, it’s difficult to create new forms of networks. So, what we’re focusing on is: We’re finding out what applications will be required next. The fact that we’re focusing on that question is a big difference between our research and that of other groups.”

Currently, the Kaneko Lab is doing R&D on ways to make it easier to handle high-resolution broadband video data.

The Kaneko Lab is developing “Content Espresso,” a wide-area distributed storage system that distributes the storage of large files among servers worldwide, and “Catalogue,” an autonomous distributed file relationship sharing system that manages and shares the relationships between digital data on the Internet separately from the data itself. Using those systems, the researchers aim to create systems that can be applied in current network environments, by developing a system for receiving and playing broadband video such as 4K and ways to efficiently utilize playing environment archives for digital content.

Also, in the MoSaIC Project at the DMC Research Center, which utilizes Keio University’s all-round capabilities, the Kaneko Lab is creating “catalogues,” using simple graph structures, for digital representation of relationships between digital data. By superposing and combining multiple catalogues, the aim is to digitally represent the relationships between data and the diversity of context.

“The relationships we create here are not those described in natural language so far; they’re represented in extremely simplified forms that are easier for computers to understand. So, a computer can smoothly understand that “there’s a relationship between this content and that content.” In that case, a search result that used 100 units of energy to produce could take just 0.1 units; it’ll be possible to obtain the same result much more simply, at far lower cost. If so, the remaining 99-plus units of energy could be used for additional processing.”
“We aim to create network technology that enables computers to use networks to help computers, enabling computers to do higher-level processing. We think that could enable computers to have more beneficial effects in daily life.”

свернуть