
Gigabit Network research and development project sub leader
Kochi University of Technology (Head of Department of Information Systems Engineering, Information Systems Engineering Course Chief) Professor Kazunori Shimamura |
Gigabit Network Research and Development project assistant leader.
Kochi University of Technology, Professor Kazunori Shimamura
Interview Conducted: 2003 April 24th
At: Kochi Communication Traffic Research Center |
Introduction
We interviewed Prof. Kazunori Shimamura, of Kochi University of Technology, who is engaged in research as an assistant leader of Kochi Communication Traffic Research Center, and some researchers such as Dr. Kanji Kato, Dr. Kisyo Takamatsu, Dr. Hideki Hayashi, Dr. Toshikatsu Kanda research fellow, Dr. Yutaka Kikuchi research fellow ( an assistant professor of Kochi University of Technology) and Dr. Takuji Nakahira (who plans to be a research fellow).
|

Upper stand from left Takamatsu research worker (Ph.D) , Kanda fellow (Ph.D), Professor Shimamura, Hayashi research worker (Ph.D). Lower stand from left Nakahira (Ph.D), Ms. Kusunose as a clerical assistant and Kato research worker (Ph.D). |
|
|
Please tell us the establishment of Kochi Communication Traffic Research Center of TAO (hereinafter referred as Kochi RC)
(Shimamura) As Professor Aoyama of Tokyo University, who is a general sub leader of the direct research of JGN mentioned, to start the construction of test bed called JGN, how next generation internet can be conceptualized and developed was the framework of TAO's direct research. Kochi RC was established at the same time when Makuhari Gigabit Research Center (hereinafter as Makuhari RC) was established. Utilizing large-scale network like JGN and to conduct research, establishment of some research base far from Tokyo was necessary. Since Kochi RC is installed inside of the Kochi University of Engineering, research is conducted under the cooperation with universities and business enterprises.
In brief, please tell us what kind of research is undertaken in Kochi RC.
(Shimamura) Each researcher will explain details of the research. We conduct various research such as P2P type shared virtual space connection (Kato), Research and development of packet assembly (Kanda research fellow), Network control (Hayashi research worker ), Sophisticated TV on demand and database distribution (Nakahira), Next generation network interface card (NIC) technology (Takamatsu research worker ). Among limited researchers, we mainly focus on next generation application research and QoS research, which we have been creating good results. |
(click to enlarge)

(click to enlarge)

(click to enlarge) |
Let me interview about P2P type shared virtual space connection by Kato research worker.
(Kato)This research relates to virtual reality space system that is to browse various information in the deep space created by CG that allow users to talk with other users using sound and images. In the future when high-speed network penetrates, each user is expected to have a personal space and send information to talk with other users using that persona space.
Quite simply, we are aiming a realization of home space, which is a space form of web sites. In order to create enormous numbers of personal space, mutual space links, that is to say link method was considered to be the key. So far, flip a page method has been used for mutual space links as it has been used for web sites. However, the space is shared by a huge percentage of the users who can move with 3D avatar personification and information that changes as time goes on by talking with other users.
We thought of these issues and realized a new link method that enables us to move to link space easily and smoothly by obtaining real time information. That new link system can be compared to Doraemon's (Japanese mouse robot cartoon character) anywhere door (This door can take people to anywhere they wishes to go.)
It is possible to install "the door" which links to any space you wish to go to any place in the space and through the "door", you can see the scenery that the link represents and by walking through it, you can easily move to the space the link represents. This link system is verified as a function element that connects to space mutually by duplex. From now on, I plan to connect users' end-to-end to peer to peer on the network so a number of people can share each personal space mutually. , And I would like to finish this up in order to construct a collaborative virtual space system whose links mutually interact.
(Shimamura) This research started under a prospective that this research will be a future web site. We thought of full media after multimedia. Furthermore, regarding full media, taking advantage of the CG was taken for granted.
|
|
In the next place, please comment on research regarding QoS by Kanda research fellow.
(Kanda) Simply speaking, this research is to piece together scraps of packets flowing in IP to one packet once again in order to diminish router's exchange loading. With current Internet technology, although packet segmentation function is implemented, packet assembly function like I developed with students of the Kochi University of Technology is not implemented yet.
For example, Ethernet can transmit packets dividing to 1,500 byte and quite many of their sizes are less than 1,500 byte. Furthermore, a core network, which can deal with much larger packet size will increase its loading towards processor as it confirms the destination of the each packet that has diminished. The increase of traffic is much larger than that of processor performance, so technological development of decreaseing processor's loading is expected to be more important in the future.
We developed PASTA (Packet Assembly System by TAO) and this is expected to demonstrate more effect with Internet Exchange (IX) vicinity and core network. And a packet such as VoIP(Voice over IP)which requires high real time implements technology that swiftly passes straight through rather than assembles Similar research can be seen in other universities but TAO's direct research enabled us to reach commercial product level. This product was already applied for patent. |
|
 |
|
If my memory is correct, Kanda research fellow received a prize.
(Shimamura) He won CQR Contribution Award Prize at International Conference of IEEE in 2001.
Kanda research fellow has been a researcher from the beginning of JGN operation till October of 2002, hasn't he?
(Kanda) Yes. Currently PASTA is being upgraded and I continue to conduct the research. In the context of technology to label the packet, it is similar to MPLS(Multi Protocol Label Switching). However, MPLS is lower layers technology than IP and the research I conduct is IP technology, so they are basically different. Since this is quite an interesting technology, I consider to suggest this to international standardization but honestly speaking, I am to busy to do that. |
|
The research theme of Hayashi research worker is also QoS. Please tell us about "Network performance improvement by self-suppressive type flow control."
(Hayashi) What I research is same as what Kanda research fellow does _ - QoS but its approach is different. Network research is divided into transmission technology research ( research of lines) and node technology research.(research of spot). As for transmission technology, it has developed focusing on the optical fiver technology. On the other hand, as for node technology, it has developed remarkably as the emergence of gigabit router exemplifies, but this advancement also led a coexistence situation of high speed and low speed node in the network. It is ideal to structure network with high-speed router only but it does not mean that lowspeed routers will disappear all of a sudden in practice. A big problem is in the case that lowspeed nodes are sandwiched by high-speed node, so a sender will send a large number of packets by seeing the final specification of the receiver only.
My proposal is that senders will send an embedded logic that takes into account slight consideration of neighbors' situation in addition to final specification of receivers Since I obtained good results from simulation, I am preparing some verification tests. I also applied a patent for this as "An auto suppressive type flow control method." |
There is a studio facility in Kochi RC.
(Click the image to zoom) |
Could you explain what do you mean by Sophisticated TV on demand and database distribution ?
(Shimamura) Mr. Nakahira, who tackles with this research, is a doctor at Kochi University of Technology and I would like him to collaborate with us as a research fellow. Kochi prefecture had only 2 kinds of channels for commercial television up till very recently. Therefore, we are thinking to meet a need to combine programs broadcasted in Tokyo and deliver them. Since Makuhari RC and Kochi RC have studio facility, I was thinking to utilize existing facility by cooperating with staff at Kochi broadcasting stations. So far, we haven't reached joint research but our research activity has been drawing their attention keenly.
(Nakahira) Currently, I carry out research investigations regarding distributed database, which forms the basis for TV on demand. This research field is called contents delivery network (CDN), which cuts across network and storage. Current animation delivery decreases delivery quality when network congests but by delivering it from the contents that are dispersed on the ground of the network, its quality is maintained. When the quality decreases in the middle of the delivery, other servers' QoS pathways and end users that have copies can be checked and if there is a path that can maintain the original contents' quality, the server can conduct the subsequent delivery. Or the idea is that multiple servers can cooperate and share each delivery that is suitable to each QoS path and rate so the original quality of the contents are offered to users without degradation. |
|
What about the research theme of Takamatsu research worker? What is "Research on data -driven distributed process system that make full use of the ultra high-speed communication transform structure." like?
(Shimamura) To put it simple, you can think it as a research and development of Next Generation Network Interface Card (NIC). I strive to develop something that allows high quality animation communication by connecting video camera to its terminal.
(Takamatsu)Data-driven type processor (DDP) is a sort of a parallel computing and its concept contradicts to commonly used Noiman type processor (It stores program as data in memory device and its method is to read the data in order to executed them). DDP embeds its orders in data so it minimizes the overhead that relates to parallel processing. As for Noiman type processor, multi processors are used for parallel processing via pipe line processing in general and Pipeline Stall (pipeline is stacked) is inevitable at any price while DDP doesn't generate this phenomenon. With animation transfer processing that I deal with as my research target, we can see the performance difference depending on how efficiently parallel processing can be conducted. Presently compare to MPEG delivery form, which is frequently used for animation transmission, the feature of DDP is its good response.
|
|
 |
|
What is your self-evaluation of the research and development activity in Kochi RC?
(Shimamura) As a research base for JGN, we were able to promote QoS research in broadband era, and we could focus on application research. Apart from PASTA and several patents, I would like to make a comprehensive survey of the research over the final one year. Tohoku detached lab shows considerable achievements in terms of network measurement research and implementation collaboration with Tohoku University. So establishment of the Kochi RC in the campus of Kochi University of Technology was beneficial. It is because for researchers who are already members of society, interchange with university students are stimulating and this also contributes to promote research partnership regarding thesis and dissertation.
Also, a number of visitors have toured Kochi RC so far. This has contributed to the expansion of JGN utilization , hasn't it?
How about the oncoming research and development of test bed expectation? ?
(Shimamura) In the case of JGN, research core was constructed from direct research and it expanded through public offering research and general utilization , so this demonstrates well function of the JGN as test bed. As for network research, it is important to proceed it utilizing something function in reality so after completion of the JGN project, I expect it to be taken over to the next research and development test bed at its best. Centering Professor Esaki, IPv6 research on JGN should be remained as it is highly evaluated both nationally and internationally. |
|
By the way, isn't Kochi University of Technology a very beautiful place?
(Kikuchi)I hear that National and Prefectural universities have a large numbers of regulations for each building but since this is a private university, we maintain and stick to the beauty of campus considerably.
(Shimamura) Kochi University of Technology is a public run private university, which was thought to be a Prefectural Engineering University at first. However, considering the free management, private university form was adopted in reality. Prefectural governer, Mr. Daijiro Hasimoto serves as a head director of the university per se and even after the establishment, finance support to the university from Kochi Prefecture continues. Following The University of Aizu, this is the second publicly run privatized form across Japan. By seeing our successful management, many others appear to follow us. Since the core industry of Kochi prefecture is prime industry, its industrial structure is old so to create a new industry, Kochi has been tacking with IT earnestly above all. |
|
Closing
I would like to extend my gratitude to Professor Shimamura,who spared more than half of the day to introduce the research contents as well as facility and I am extremely grateful to other staff ,too. According to a clerical assistant Ms. Kusunose, Kochi University of Technology where Kochi RC is located was originally an experimental forestry station so various unique trees are planted. Since campus is so beautiful many couples who had wedding in Kochi city visit to the university in order to take some souvenir shots. If you haven't visited here yet, I recommend visiting here.
The full responsibility for the content lies with : JGN Web Editorial Department |
JGN Web Editorial Department seeks JGN project researchers who can cooperate with an interview. Please let us know your name, e-mail address, telephone number and JGN project number and information such as self- recommendation/nomination to the jgnweb@jgn.tao.go.jp. |
|
|
 |