Industry View: Hideo Kuwahara
Dr. Hideo Kuwahara leads a research group for long-range communications at Fujitsu. There he has retained the responsibility for overseeing photonic research activities from his former position as the director and general manager of the Fujitsu Laboratories. In 2006 he was named a Fujitsu Fellow.
“I was responsible for the research itself – I managed the organization, performance estimation, and budget. Now, as a Fellow, my main task is no longer to bear that kind of management responsibility. It has been shifting to coordination, advising, supervising – keeping an eye on the larger team concept.”
It is a lot to keep an eye on. Fujitsu, a company that is up to roughly 70 percent dedicated to manufacturing computers and associated services, and 20 percent to building telecommunications infrastructure, has a substantial network of laboratories, with about 1,500 staff members. The majority are located in Japan, but some parts of the network are overseas. In the period 2000–2003, Kuwahara was assigned to establish a new laboratory in Texas.
“Those were the years during which Fujitsu, like other IT companies, had to weather the bursting of the dot-com bubble. This meant a very severe downturn in 2001, with low revenues. After that came a recovery; interest in photonic technology for data transportation really returned because of increased traffic.”
“And this still continues, even at the end of 2008 and the beginning of 2009, when there was a financial meltdown. Fujitsu was not immune, but it is my impression that companies like ours did not suffer as badly as, for instance, the automobile companies. The telecom industry is about infrastructure, so we were not hit as hard. We live in times in which many new applications, such as YouTube, are hungry for bandwidth.”
However, Kuwahara warns that, in comparison with electronic technology, photonic technology is somewhat immature: “In 100 Gbps systems, we would like to use the coding methods now used in wireless networks, called quadrature amplitude modulation (QAM) and quadrature phase-shift keying (QPSK). In that respect, even if it works at lower speeds, wireless is more advanced.”
But bringing these coding techniques to networks where light is the information carrier is not so simple – not if, at the end points, the light is to be converted into electrical current and dealt with by electronic components. Kuwahara explains this as follows: “For 40 and 100 Gbps, the electronics work too slowly to keep up. Recently, complementary metal–oxide–semiconductor (CMOS) technology has been increasing in speed: data rates as high as 40 Gbps can now be reached, we think. In that case, it would be a natural trend to keep combining optics and electronics. The electronics would provide fast processing capability, and the optics would give us fast transfer rates. But there is a limit as to how far you can push this. Fujitsu is a partner in a 100 Gbps facility; a lot of companies are working on networks with channels of that speed. Electronics will not be fast enough. With QPSK modulation and CMOS components, around 50 Gbps would seem the limit for electronics right now.”
Merging Electronics and Photonics
One way to keep electronics in play is by keeping the data rate for each channel lower than it could be. This allows for the spacing of the channels closer to each other with respect to the wavelength, thus increasing the number of channels available.
Kuwahara: “Another way to circumvent the limitations of electronics would be polarization multiplexing. In every channel you could send two light signals with different polarizations, so they are independent. Keeping the channel spacing the same, we could reduce the speed per polarization direction by half; with two polarization subchannels of 25 Gbps, CMOS technology can definitely be used.”
Another glimmer of hope is the advent of new electronic materials, promising faster electronics. Kuwahara agrees: “There is indium phosphide, which could be combined with silicon and germanium into compound devices. But the scale at which this can be done is very limited. It cannot be done as CMOS.”
But how far will CMOS be able to bring us? Advances in CMOS technology will mean, by definition, increasing the switching speed while reducing power consumption. According to Kuwahara, “this means that we will have to push miniaturization forward. But we are already getting near Moore's Barrier, the level of miniaturization below which Moore's Law doesn't apply any more, because the size of atoms becomes appreciable and influences the functioning of devices. This is where optics could take over or may even be one of the few alternative solutions.”
“We will then have to look for completely new devices that have not yet been invented. Optical processing has the capacity to be fast enough. At high bit rates, processing should be only optical. Work on this is being done at several places, notably at the Heinrich Hertz Institute in Berlin, and also COBRA in Eindhoven. But that kind of activity is still at an early research stage – there is still a lot of fundamental research to be done.”
Devices can process light in vastly different ways. One way that Dr. Kuwahara thinks might be a good candidate for development is through the use of nonlinear glass fiber. In such a fiber the transmission properties change as the intensity goes up. “At Fujitsu we are using this kind of fiber to obtain a noise-reduction effect.”
Another very important technology is already in practical use: wavelength-selective switching (WSS). It is used in ROADMs – the switches that allow access to and exit from the glass-fiber highway. “The reconfiguring means the direction of traffic can be changed. They work very slowly now, but in the future we should be able to obtain faster redirection until, in highway terminology, you can switch truck by truck, car by car. But current technology cannot do that yet.”
Can we be confident that these technologies will be ready to help us cope with traffic demand in ten years’ time? “Partially. Optical technology is now advancing. It is becoming more device-directed. From silicon crystals with some electronic as well as photonic properties, we progress to photonic crystals, materials with a periodic structure to give us the special performance we want. With this, you can make highly integrated, very sophisticated small devices.”
Telecommunications Are Changing
Of course, things may turn out differently. Unexpected technical roadblocks might appear, but progress might also be hampered by economic and even legal developments. “Telecommunications are changing. At first, you had just the telephone, but that's less important now. Instead you have e-mail and video. And those are now starting to merge with broadcasting activities. But currently we are unsure: What is the direction all this will take? How should we manage all this data? With regard to copyrights, for instance. That's the kind of border we are at now: are we communicating, or publishing?”
“The regulations on these topics are not maturing at the same rate in Japan, the US, and Europe. The business models are not so well established. Yet these will be decisive.”
Fujitsu Laboratories LTD
Vice General Manager
1-1 Kamikodanaka 4 chome
Company mail No. L50