We interviewed Prof. Tan Yasuo, of Japan Advanced Institute of Science and Technology, (JAIST), who has been approaching the integration of the network of electric appliances and the Internet earlier on. He developed and merchandized his own product and it was a great success. We asked him how research that would contribute to the real world could be produced.

Coincide with the timing of the JGN development, not only network researchers but a large number of users such as medical experts and educators widely used the Link Unit Under favor of the timing, I had a chance to be famous as "Prof. Tan of a Link Unit."

What was the concept of the research?

I thought of connecting the following 3 networks interchangeably: computing machine network, AV device and home electronic network, and ISDN telecommunication network. As for, electric household network for example, it has been researched by echo net group.

Recently, there is an argument that everything should be connected to IP, but what I thought was a little different from that. I thought the gateway should be where the network is interconnected and the protocol would be converted there.

You can imagine that the protocol IEEE 1394 used for electrical household is widened via ATM. In fact, only DV flows there and IEEE 1394 network over ATM doesn't really exist there.

Why was the Link Unit accepted so widely?

One is the limitation of its function. As an engineer, I expected to install various functions to Link Unit, but daringly I limited its function only to tele-conference. By limiting its function, I think users were able to accept ATM for the first time.

The other is the large impact of receiving and sending animations in 30Mbps band. Compare to receiving and sending streaming animations over the Internet in a narrower band, I thought its overwhelming quality was sensational to everybody so it greatly attributed to the success.

Are there anything that you have noticed through Link Unit?

As commercial goods, they did fairy well. . Thanks to JGN, Link Unit was accepted and due to Link Unit, people utilized JGN and there were some people who thought of utilizing JGN.

As for me it is important to make things routine that are already taken for granted. I think Japanese are relatively not good at doing this. Technologically speaking, both ATM packet and IEEE1394 communicate via fixed-length packet and many researchers lose interests when they take it - that communication via fixed-length - as a common place. In fact, when you actually conduct it, various elements hinder its communication occasionally. Therefore, it's important to build up efforts of those already taken for granted logically in actual manner. And to do something that someone else has to do like this is an extremely important issue.

What are the difficulties of this research?

Since this research was collaboration with a company, dealing with intellectual property right came into question. JAIST is a national university, and with a use of a standard contract, it would be quite advantageous to the nation since its information would be siphoned off to the government. For companies aspiring worldwide business with intellectual property right as a core, incentive of conducting collaborative research would decrease , so this issue was quite tough to adjust.

Another difficult point was some speculation differences in a company depending on engineers, the whole company itself and sales and marketing. For example, regarding Link Unit, some engineers raise ideas to release much low-cost products while the sales side expects profits. Therefore, with slight expectation of profitability, this idea disappeared.

However, it is more worthwhile to release an actual product and acquire users by cooperating with a company, than the troublesome described.

Compare to the first test mode (picture on right), the commercial product ( the picture left) became much thinner. This realized due to the collaboration and effort of the company.

How did the collaborative research start?

When I spoke about the concept of Link Unit to a person at Sony Corp, I was introduced people who have similar concern. Just at that time, there was a team that was thinking to flow ATM packets on IEEE 1394 inside of Sony Corp, (actual appearance of the Link Unit is different from this) and we decided to proceed research together since our area of interest was close.

What sorts of themes are you researching currently?

Currently, I am participating in a project called Viz Grid (Collaborative Visualization Environment on Grid Computing ) and I am conducting research that aims integration of virtual reality and the real world. I am in charge of network related research in this project. In this research, there is a system called "agency terminal" and this can project participants' visages of remote participants on the monitor. It enables remote participants to operate camera angles and direction of monitors on his/her own . It gives feeling as if you are moving your own eyes and neck. It is different from simple TV conference and when you use the "agent terminal", existence of the remote participants greatly increase. When you use the test model in reality , interesting enough, the larger the screen becomes, the more influential that the "agent terminal" becomes.

By the way, how do you evaluate JGN?

Firstly, it is an extremely worthwhile place to conduct incubation. Link Unit is one of the successful examples, but it is not easy to commercialize products from academic research and transfer it to penetration phase. In spite of active research , there is "dead valley" problem. In fact, I feel that there is no special formula to out step this difficult point called" dead valley." Yet one important point is what I mentioned earlier, to take things for granted that are already taken for granted.

Secondly, due to the realization of JGN, human resources that are capable of using wideband network definitely increased. Apart from partial organizations whose budget is abundant in metropolitan areas, this chance couldn't carry out without the existence of JGN. Especially at local regions, significance of JGN is more absolute than that in metropolitan areas.

Furthermore, another effectiveness is the increase of people who can realize and understand the benefit of using high-speed network. You can feel the flow differences between a few small-sized kbps streaming and tens of normal DV kbps. Many of you have seen the DV class images in the symposium like event and some of you might supported the events . These who experienced high-speed networkcan't stand without using it. . By offering stable leased line service, installation of access points throughout Japan and realization of wideband high-speed network environments is really significant. I think elimination of access links to access point regardless the distance and free use of circuit also contributed to increase its users.

Finally, Gigabit Center and Network Operation Center's (ONC) clearly improved information desk for inquiry is assessable. . Even with a large-scale research project, there would be a case to close its inquiry at some famous professors' place and resume its implementation. In the case of the network test bed, its openness is worth to various people. The meaning of faire and open operation of the test bed by Gigabit Center and NOC is very significant in my opinion.

JGNv6 has been operated as auto-decentralized network by researchers but does it mean that the conventional operation is preferred?

I am involved in JGNv6 as an operation administrator and regarding JGNv6, and I think current community base research system is preferable. It is because IPv6 itself is the technology that requires further research form now on.

However, generally speaking, it is difficult to state whether the community base or businesslike base is better. It is because for the internal members of the community, the productivity could be high, but for the outsiders potential utilization opportunity might be lost as there are too many unclear parts. There are not many people who point out this point. Well, at any rate, openness of the JGN's information desk for inquiry should be highly evaluated.

How about expectations for next generation test bed?

I expect test bed that enables to research (light wave length) lambda by (NTT link) access. This is already introduced in Canada and I think it is good to have lambda network that travel throughout Japan.

Canarie of Canada

In JAIST, there is a project called AYAME that implements MPLS router of their own making and this team starts to response to lambda. This also promotes research and development using JGN.

AYAME

Since I am interested in the edge part of the network , I expect an emergence of a new research environment.

 

Finally, are there any comments via network research on a routine basis?

For the career of a researcher, writing thesis is the top priority and there is a mechanism that practical creation is still difficult to be evaluated. I often hear stories that researchers of popular open source software had to repeat a year of the master's degree since he/she was developing it. Since creating something practical is as important as writhing thesis, I think system that can evaluate practical creation is also vital.

I strongly felt that Prof. Tan, who researches network interconnection apart from IP protocol, does have ability to connect different cultures.

I am grateful for Dr. Tan's interview out of his hectic schedule at the end of the year.

The full responsibility for the content lies with : JGN Web Editorial Department

Echo net
Canarie Canarie in Canada
AYAME