Tokyo

Friday morning, after a breakfast of green beans, cocktail franks, and a salad, I met Kenji Naemura, at the time a vice president of Nippon Telegraph and Telephone who has since become a professor at Keio University. We sat in the back of a car with shiny white seat covers and drove 50 km south of Tokyo to NTT's Yokosuka research laboratories.

Naemura began his career in the 1960s in Tokyo using a machine developed by NTT that preserved the Illiac I interface but had a different architecture, possibly the world's first clone. Naemura developed the hardware architecture for NTT switches, then spent two years at Champaign-Urbana working on the Illiac IV.

When he got back to Japan, the focus moved from switch architectures to communications software, and Naemura began playing a central role in the Japanese delegation to ISO. He was involved in the very first OSI meetings where the seven-layer Reference Model was developed. Although the Japanese delegation would not have been unhappy to have been referred to as the seven samurai, Naemura felt that perhaps the architecture should have a less rigid stratification of the protocol architecture.

Naemura was bringing me to Yokosuka to learn more about NTT's vision of what the telephone network would look like in the next 25 years. Dubbed VI&P, for "Visual, Intelligent, and Personal," the vision had been developed over several years as a broad consensual effort within the corporation. The previous shared vision had been the installation of narrowband ISDN around the country, and with that effort well underway, NTT, the largest corporation in the world, wanted to give its employees a sense of common purpose.

Basically, VI&P outlined a B-ISDN, all-fiber telephone network with 620 Mbps channels into the home and one or more terabits of capacity inside the network. NTT was targeting a 15 year development effort, with large-scale deployment in the early years of the 21st century.

Much of the VI&P effort dovetailed into current products. For example, NTT had recently announced a 230 gram portable telephone. One of the VI&P goals is for a 50-gram personal phone, coupled with an intelligent network to find the phone wherever the person might be.

We started our tour by looking at some of the prototypes, developed for press briefings, management demos, and other shows. Although there were a half-dozen systems, the one that I found the most interesting was called "teleview."

The heart of teleview was image processing software to extract objects from a video feed. The prototype had a camera trained on an NTT office and used ISDN to send the video to a Yokosuka workstation. Based on movement of people and the location of desks, the computer could determine if people were sitting at the desks.

My immediate reaction was that this was highly Orwellian, allowing supervisors to track the length of bathroom breaks. While toilet tracking was certainly one of the possible applications, there were others.

For example, a touch screen on a workstation had a grid, with one square for each employee. If the employee was at the desk, the square was green. Touching the square automatically dialed the telephone. Of course, if a dog were sitting in the employee's chair, the square would also be green, but the phone might not get answered.

We touched a button and a poor harassed employee of NTT ignored it for a while, then finally picked it up.

"Just testing, thanks a lot," my tour guide said. The employee didn't bother turning around and I got the impression that he wasn't as thrilled with teleview as the researchers were.

The marketing video for VI&P showed a couple of other teleview applications. Applied to a parking lot, teleview could direct drivers directly to empty spaces. At an amusement park, the system could count the number of people in line, displaying the expected delay for each ride on monitors.

After a walk through various labs, looking at ATM switches, HDTV systems being fed with 150 Mbps lines, and prototype equipment for distributing fiber from poles to homes, we drove back to Tokyo to NTT's Kasumigaseki showroom to look at ISDN equipment.

Japan has very aggressively deployed narrowband ISDN, providing service in two thousand service areas distributed over all the islands. Both basic rate and primary interfaces are available, and I met several people with basic rate service in their homes.

The showroom was full of slick equipment ranging from G4 digital fax to dedicated floppy disk transfer devices. Videoconferencing stations, ISDN pay phones, and portable video phones were all available. Although the cost for equipment and service is not cheap, NTT had managed to attract over 60,000 basic rate lines by the end of 1991.

Throughout my visit, people kept pointing to picture phones, and it was obvious that they hoped this device would spur personal ISDN use. While picture phones didn't appear to me to be the Lotus 1-2-3 of the ISDN market, I did see one device that had real promise, the ISDN Karaoke Machine.

Using 2 64-kbps B channels, a large database server could send audio down one pipe and a video image with song lyrics and a bouncing ball down the other. The system kept the bouncing ball synchronized with the audio track. Here was a way to turn every room into a karaoke bar.

Dr. Naemura dropped me off at NTT's R&D headquarters where I met the three scientists who were translating Stacks into Japanese. Shigeki Goto, Ken-Ichiro Murakami, and Hisao Nojima came walking into the conference room armed with huge stacks of paper. For the next two hours, we went through a long list of what they politely referred to as "questions."

Most of the "questions," of course, were examples of typos, miscalculations, and other errors on my part. Not only had they scrutinized my text on a word-by-word basis, but in most cases had gone back to the original protocol specifications to check all data. It occurred to me that perhaps we should translate the Japanese version back into English for the next edition.


Saturday, Jun Murai had asked me to come to the monthly WIDE meeting. With a few hours to kill before the meeting, I went to the Akihabara "Electric Town" to look at the latest palmtop computers.

An intriguing discovery was the Takeru Club, a vending machine for selling software. In the center of this 4-foot high machine was a touch screen to select software. On the right were receptacles for bills and coins. On the left were floppy drives for three different diskette formats, and even a slot for dispensing laser discs.

Restraining myself from buying software, and tearing myself away from a 14-channel programmable controller, I caught a series of trains down to the Yagami campus of Keio University. I got a can of hot coffee from the vending machine and sat down to wait for Jun.

When we arrived at Keio, we found a room full of 40 volunteers, students, and scientists, hard at work discussing internetworking issues. Research results were presented, the latest routing tables were discussed, and new plans for WIDE were formulated. This group gathered once a month for an all-day Saturday session and had previously met twice a month. Monthly meetings by these volunteers are supplemented every year at two camps, where all 65 WIDE researchers and students collect in the middle of Mt. Fuji, wiring themselves into the network with Ethernets and ISDN.

I gave my lecture, a rambling discourse on resource discovery, then read my mail and went back to Tokyo and wandered around the Ginza entertainment district. In a liquor store, I decided against buying a bottle of Lemon Pie and Cream and went next door to a random sushi joint for a dinner of uni and kazunoko. On the way back to the hotel, I stopped into the Atlantis Blue bar, a fairly new establishment proudly sporting "Since 19XX" on its sign, figuring they deserved my money for using wild cards on their date.


Sunday morning, suffering from newspaper withdrawal, I went in search of the International Herald Tribune. In Japanese, one asks for the paper as "the Geraldo," but a search of a half-dozen hotels in my area yielded nothing.

After two hours of walking, I was getting hungry, so I followed the railroad tracks to the next station. Under the tracks was a little sushi restaurant, no bigger than your average closet. A dozen stools were set around a circular counter. In the 2 by 3 foot space in the middle of the counter was the chef. A conveyer belt full of plates circled this island and customers simply grabbed plates as they came by.

Over the belt was another level which held tea glasses, chopsticks, jars of radish, and, in front of each stool, a spout for dispensing hot water, and a bucket of tea bags. If you didn't see what you wanted, you shouted out and the chef would make it and hand it out. After a half-dozen dishes, I forked over 480 yen (U.S. $3.84), amazingly cheap for Tokyo.

Fortified with octopus, I spied a bookstore, found my Geraldo, and went back to the hotel to await my dinner appointment that night at Jun Murai's house.

Jun lives in Seijogakuen, the Japanese Hollywood. I boarded the wrong train and ended up in Sagami Ohno, the equivalent of going across Manhattan by way of New Jersey. The trains were full of commuters, many wearing white surgical-style face masks worn when sick to prevent the spread of disease.

After a few more wrong turns, I finally ended up at Jun's house. The house, on the same plot with his parents, is unusual by Japanese standards. Apart from its large size, it features built-in Ethernet cabling, wiring for 5-channel Dolby stereo, and a truly impressive area of electronics gear. A little table near the entrance held a stack of 12 remote controls for the VCRs, televisions, CD players, and other evidence of many, many trips to Akihabara.

Over a dinner of salsa and margaritas, smoked salmon wrapped in horseradish and California wine, and roast pork with champagne, we talked about–no surprise here–computers. Jun keeps his finger right on the pulse of the latest technology and seemed to already have used most devices that I had only heard about.

Monday morning, I found a Yoshinoya shop and had a bowl of fatty beef over rice garnished with a raw egg. Totally enamored by this time with Japanese food and electronics, I pulled out the Japan Times to look for a place to live. A three bedroom apartment–"with telephone"–was available for a mere 1 million yen (U.S. $8,000) per month, but the six month deposit would have put a real crimp in my hardware acquisition plans. A nice room of 42.05 square meters was only 200,000 yen, but I wasn't quite sure how I could fit my computers into a space that small, let alone find room for a bed.

At 9 A.M., I met Tomoo Okada from Fujitsu for a trip down to the new Tokyo city hall, at 60 billion yen (U.S. $500 million) reputed to be the most expensive building ever built. We took the train to Shinjuku station, which handles a mind-boggling 3 million passengers per day, and walked out the west gate to the city hall.

The complex consists of two towers, 34 and 48 stories high, and a 7-story city council hall, all clustered around a 5,000-square-meter outdoor Citizen's Plaza. All the buildings are made of a lattice of light and dark granite, which, along with the reflective glass makes the complex look like a huge integrated circuit.

We started on the first floor in the main control room for the large tower. The room is divided into regions, one for each of the major subsystems. Each region has a series of video screens and keyboards. The elevator panel, for example, shows the status of all 20 elevators and has emergency shutoff switches for each one below each screen.

Another set of screens shows the status of all 800 card readers in the complex. All employees have magnetic cards, and these are used to clock in in the morning, to open conference rooms, and even to start up computers. At the far end of the room, a technician was using a mouse to click on the diagram of one of the floors. He selected a room, and with his mouse adjusted the humidity setting from 40 to 50 percent.

We took an elevator, equipped with extra buttons at waist level for handicapped access, up two floors to a public videotex room. Here, citizens can pull up and view or print demographic data, schedules of cultural events, and city council meetings, as well as a wide variety of other types of information. That same floor features a publications sales outlet, a citizen counseling area, and other public services.

On the 13th floor, we entered an office area 108 meters long housing part of the MIS group. This floor, like all the other office floors, has an "OA Room," which houses PCs for secretaries as well as the wiring closet.

The complex features an optical backbone with 7 km of fiber running FDDI. Access to this backbone is through 100 bridges located in office complexes, machine rooms, and other locations. Attached to the bridge is a star coupler which is used to distribute fiber out to the workstations. These local loops run Ethernet and add another 295 km of fiber.

On the 9th floor, we looked down from the viewer's gallery into the 2-story room housing the Disaster Prevention Center. This control room is used in the event of natural disasters, such as earthquakes, typhoons, or large fires.

In the middle of the room are two 200-inch video screens. Video feeds from helicopters and other sources can be displayed on the screens, as well as computer simulations which show the expected spread of fires. On either side of the screens are other displays that show the current weather conditions, the reporting status of civil disaster teams, and other information needed to determine strategy.

Next, we went up to the 10th floor, devoted to seven 3090-class mainframes from Fujitsu, Hitachi, and IBM. Another machine room in the other tower has another five large systems, for a total of 3,000 square meters of machine rooms.

Each machine room has a fault-resilient IBM System 8B, featuring dual processors, dual disk drives, and dual buses. The two machines act as hot backups for each other, providing a second level of protection. The System B8s are used for environmental monitoring, using a token ring to control and read information such as power status, humidity, and temperature. In case of an exception condition, the environmental monitoring system could initiate an orderly shutdown, dial telephones for system programmers, and sound alarms.

Each of the machine rooms features a special earthquake-resistant raised floor. In case of large shocks, the steel lattice will absorb and dampen vibrations, rolling the entire machine room in the opposite direction from the building. Impressive as all these systems were, I was most intrigued by the analog data jack on the ISDN pay phone in the lobby. The next morning, I called up Jun Murai and asked him to set me up a local login, then headed back to Shinjuku.

I set up my office in the lobby, balancing my fax modem and laptop on the counter. Ten minutes later, I was reading my mail in Colorado, using the talk program to bother my systems manager, and even Kermiting my files down to my laptop to work on later.

Periodically, people would sidle over to try and figure out what this crazy Gaijin was up to. Each time I spotted somebody looking over my shoulder, I would point excitedly at the screen and say "America!" This was usually enough to evoke a polite smile and a hasty retreat.

At Narita airport, I set up my laptop in the lounge to work on the files I had downloaded. Sipping a scotch and munching dried squid, I noticed that people kept their distance.

<<<>>>