Research Activities in Ultrafast Optical Logic Lab, Univ. Electro-Communications, Tokyo
Titles: English (upper) / Japanese (lower)
Texts and captions: English
Since: July 5th, 2006
Last updated: September 20, 2006

Research Activities in Year 2006
Ueno Laboratory, UEC, Japan

Real-time Teraherz communication technology
with photonic circuits
(1012 Hz= 100 times more of present frequency in standard 2×1014-Hz band)



On-going targets in year 2006

Building experimental prototype circuits, for use in 160-GHz, 160-Gb/s all-optical gatings research
160GHz, 160Gb/sの全光ゲート実験回路の製作
We started new research activities in this university in 2002. In our second stage since 2005, we are more intensively building stronger experimental setups for use in our all-optical gatings research. Our target frequency zones in the second stages are (1) 40-160 GHz and (2) 100-400 GHz. voice

For generating these ultrahigh-frequency optical input signals, we are hand-making (1) low-dispersion optical-fiber amplifiers and (2) experimental optical-fiber multiplexers, with using one commercial, 10-GHz, synchronously-mode-locked fiber laser. The width of pulses are controlled to less than 2 ps after careful dispersion compensation and pulse compression, as needed. voice

For precisely monitoring the ultrafast optical pulses from our all-optical gates in both time and frequency domains, we are using a cross-correlator (time resolution< 500 fs, dynamic range> 103), optical spectrum analyzers (wavelength resolution= 10 pm= 1.3×109 Hz in the optical frequency range near 2×1014 Hz), and EO-converted frequency spectrum analyzers with hand-making optical heterodyne circuits (frequency resolution= 1Hz with Rb-stabilized clock). voice

Regarding more system-oriented monitoring systems such as bit-error-rate measurements, we are on our way. At present, a commercial electronic pulse-pattern generator (max word length= 231-1, 12.5 Gb/s), and some EO and OE converters are installed in our setups.

Both commercial and custom-designed SOA's, for building the ultrafast driver parts in our all-optical gate circuits
Semiconductor optical amplifiers (SOA's) are the most important physical driver parts in both experimental and theoretical research activities of ours. In the previous stage, we were using commercial SOA's from Alcatel, France (Avanex brand, at present).

In the current stage, we have started characterizing both commercial and custom-designed SOA's from active start-up companies InPhenix, USA and those from Center for Integrated Photonics (CIP), UK, and then installing these SOA's in the driver parts of our all-optical gates, such as DISC and PDSMZ-3R gate circuits.

Studying the unknown lower limits of energy consumption in our all-optical gate circuits in the 40-160 GHz frequency range, in contrast to equivalent O-E-O-converted circuits
All the way from the birth of SOA-based all-optical gate circuits in the early 1990's through this 21st century, their inherently low energy consumption has been one of the most exciting advantages, in comparison to equivalently functional OEO-converted opto-electronic schemes in the present industry. In the all-optical scheme, in fact, the record-fast 160-Gb/s wavelength converter in 2000-2002 consumes an electric power of only 0.60 Watts. In conventional opto-electronic schemes, in contrast, the energy consumption in electronic transistors and OEO converters quickly jump up from 40 GHz, due to physical circuit-dispersion limits and electron-transnport limits.
The low energy consumption in all-optical gates has been only a matter of fact, however. Nobody can tell if the energy consumption scales with frequency linearly, bi-linearly, or exponentially above a certain threshold frequency. Nobody could clearly tell how many percentage of the injected electron-hole pairs is consumed for driving the all-optical gates, either.

Now we are systematically studying the energy consumption of all-optical gates, and trying to develop its quantitatively scalable future vision as a function of frequency from 40 through beyond 160 GHz, anticipating its dependence be between linear and bi-linear (ie., power consumption, P∝ f, or, f 2) to beyond 1 THz. The all-optical gate structure that we are using often in our fundamental research is the DISC (delayed-interference signal-wavelength converter) structure, which is one the simplest structure among many of all-optical gate structures to date. (This gate structure fits well with compact experiments in universities, to some extent.)

Futhermore, We are more physically investigating the quantum-conversion efficiency from the number of injected electron-hole pairs per second to the rate of ultrafast stimulated recombination in the nonlinear SOA's that are driving our all-optical gates. We are experimentally observing a large variation in the quantum efficiency from research sample to research sample, and trying to diagnose its dependence on material and structural designs of existing SOA's. This we believe is an unexpectedly new research activity in this technical field, because we are developing the first diagnostics model, which will play a role similar to the threshold-current model expression repeatedly used in the research and development of all kinds of laser diodes.
(This research direction will be important, too, in the longer-term transition from bulk SOA's to next generation SOA's with faster nonlinear materials such as QD's and ISBT's.)

Our original, DISC-Loop-type mode-locked pulse generator circuits
In the 1990's, so many schemes of GHz-class, few-picosecond, synchronously mode-locked laser diodes (MLLD) and fiber lasers (MLFL) had been studied and demonstrated in conferences and preliminary-product exhibitions. The MLLD scheme is advantageous in terms of its much smaller footprint, when monolithically integrating such ultrahigh-frequency optical sources in optical-signal-processing systems. In our opinion, however, this MLLD scheme must rely on a-few-picosecond reponse-time constant of strongly photo-excited electron-hole pairs in 3-dimensionally-localized saturable absorber and other exceptional structures inside it, for precisely narrowing its optical pulse's waveform to below 2 picoseconds (throughout the system's lifetime of 10,000 or more hours). In the meantime the photonic integration technology prefers a limited number of elementary components, like CPU and other electronic LSI technology did in their successful history.

As an alternative option, we are studying our original, DISC-loop-type mode-locked pulse-generator scheme. Its optical pulse's waveform is narrowed to below 2 picosecond by a stable Mach-Zehnder interferometer, in principle. The DISC-loop-type pulse generator consists of an active SOA, a passive MZ interferometer (whose relative delay time stabilizes the pulse's width), a passive etalon (whose resonant frequency stabilizes the pulse's repetition rate), a passive polarization converter, and an active cw source (which stabilizes the pulse's optical frequency). Since this generator scheme consists of these elementary components and many of them are passive components, this scheme will be highly reliable in the long term and will fit well with optical circuit integration in the near future.
(Moreoever, when incorporating standard MEMS schemes, we will be able to continuously tune most of the primary specifications such as above-mentioned pulse's width, repetition rate, and optical frequency.)

measured 2.2-ps, 41-GHz mode-locked pulses, Δt • Δf= 0.53 (Rei Suzuki et al., 2006)

After achieving 2-ps, 40-GHz pulse generation in 2005, the current targets of ours are 1-ps, 40-GHz pulse generation, large transmittance of DISC converter inside the generator, single longitudinal mode, making of a carriable prototype in a NIM-standard case, etc, within a few years!

Tunable optical-spectrum-synthesizer techniques in the ultrahigh-frequency domain
We might never have a chance to ejoy synthesizing optical spectra of 100-Mb/s FTTH signals. The typical separation between their optical frequency components is just too close for us to practically separate and then synthesize.

Once we turn our eyes from 100-Mb/s to above 40-Gb/s world, it is another story, dramatically. In principle, optical spectrum synthesis should fit well with ultrafast all-optical gate circuits!
To start from, we are theoretically studying if and how we can universally design a good enough compensation for gated data waveform distortion caused by a 2nd-order relaxation-time constant, which will generally exist under faster gating conditions or in faster nonlinear materials. Up to now, we have achieved the first interesting results and are on our way to clearly wrap them up!

Teraherz-class, sub-picosecond characterization research on quantum-dot, photonic-crystal-structure wavegude samples
In the long-term nation-wide project "femtosecond-technology project" supported by NEDO in 1996-2004, for instance, nano-technology fabrication techniques for faster optical semiconductors having less-than-one picosecond response time such as quantum-dot (QD) and inter-sub-band-transition (ISBT) semiconductors had been developed (eg.: H. Nakamura et al., OSA Optics Express, Dec. 2004). Most of the experimental circuit- and system-level all-optical gate research are, however, still globally based on more reliable optical nonlinearities in bulk semiconductors, since those QD semiconductor samples are not mature enough for circuit-level gating research. We circuit-level all-optical gate research groups will make an adiabatic transition from bulk to new semiconductors in the coming 5 to 10 years in step-wise manners.

Our optical "pump-and-probe" techniques
for characterizing ultrafast temporal responses of any wave-guided optical materials:
Since mid 2005, our research group has started sub-picosecond characterization research on quantum-dot, photonic-crystal-structure wavegude samples, under a renewed nation-wide collaborative research project sponsored by NEDO. As a primary mission of ours, we characterize the saturation and resonant behaviors of the nonlinear optical-phase shift in those research samples, from fundamental research standpoints. Straightforward comparisons between nonlinear responses of these two alternatives, ie., bulk and QD semiconductors are the other valuable mission of ours, the specialist from characterization viewpoints in this technical field. We are now building an ultrabroad-band, sub-picosecond, jitter-free characterization setup in the pump-and-probe scheme, with using a strong experimental facility, 130-fs, 80-MHz parametric oscillator system in the 1.1-to-1.6-um wavelength range.

Our innovative research directions, hopes, and dreams in the coming 5 years

Designs and analyses of elementary all-optical logic-gate functions, such as wavelength convesion, polarization conversion, 3R regeneration, demux, etc. in the 160-Gb/s region
In 2000, bit-error-free all-optical wavelength conversion in the record-high-frequency 168-Gb/s, with a standard SOA in the DISC gate-circuit structure, was achieved by the research group in NEC. This frequency record was never broken for almost six years since, and then very recently another strong group in the Eindhoven Technical University achieved 320-Gb/s wavelength conversioin with an all-optical circuit structure similar to DISC (PD paper in OFC 2006).

The all-optical gated (ie., wavelength-converted) signals in these experiments, however, seem to suffer from waveform distortions, in principle of the present DISC mechanism, which had been overlooked in the champion demonstration experiments due to the normally limited time resolutions of waveform monitoring systems. Modeling research activities on these all-optical gate circuits has been much behind the world-wide experimental demonstration research activities, up to now, too.

In our previous experimental and modeling works in the fundamental research stage, at least one source and its working mechanism of such waveform distortions has clearly been figured out.

Separately, very recently, an unexpectetdly effective impact of blue-shifted spectral-band-pass filter on the gated signals was experimentally observed and then theoretically modeled well. Moreover, nobody has not taken into account other nonlinear mechanisms (nonlinear birefringence that causes polarization rotation, and carrier-cooling relaxation after non-equilibrium electron-hole recombination) in the gate-circuit models. It is very exciting for us to design new gate circuits in the second generation, making constructive uses of these ultrafast nonlinear responses for improving and accelerating these high-frequency gated waveforms.

measured and calculated sub-pulse waveforms in the gated output, respectively,
with our ultrafast cross-correlator and first-order gate-circuit model (Jun Sakaguchi et al., 2005)

schematic sketch of ultrafast interactions between input photons via excited electrons

theoretically calculated eye diagrams, according to our circuit model,
in the step-wise data-regeneration process (Masashi Toyoda, 2005)

a replica of the mankind's first electronic transistor, invented at Bell Labs,
on December 23, 1947, six decades ago.

Break-through the concerned carrier-cooling relaxation in bulk materials and other higher-order relaxations in QD materials with optical spectrum-synthesizer ciruits
In principle, the optical spectrum-synthesizer scheme fits well with our ultrafast optical signals. So does the arrayed-waveguide grating whose typical optical frequency resolution ranges from 100 GHz to 25 GHz, which had been developed as Mux/Demuxfor in dense WDM technology.

We are planning in the long term, to design optical synthesizers within all-optical gate circuits for matching bulk semicondoctors in the carrier-cooling-limited frequency region and futhermore, for matching higher-order relaxation properties of crude QD's to optical signal-processor regime in the smartest manner.

typical contrast between optical responses of a bulk-InGaAsP SOA with and without optical acceleration.
(Jun Sakaguchi, et al., 2006)
the carrier-cooling-relaxation-induced transition points have appeared in the curves.
time resolution of this cross-correlation-trace measurement was 2 ps.

Optical data buffering with optical 3R regenerator in a re-circulating loop, and even more topological nonlinear-optical logic circuit resesarch

Unexpectedly untouched issue in the lower frequency limit for our ultrabroad-band digital gatings
10年間意外と見過ごされてきた電子注入配線の低周波特性: 広帯域電子注入配線設計

High-quality, 160-GHz, 1-ps pulse generation from the DISC-Loop-type pulse generator
DISC-Loop型パルス発生回路による高品質 160GHz 1psクロックパルス列発生

Continuous tunability in pulse's width, center frequency, and repetition frequency from the DISC-Loop-type pulse generator
DISC-Loop型パルス発生回路の新展開: 連続可変パルス幅制御、連続可変中心周波数制御、超高精度光周波数コム発生

Reduction in the electric-power consumption with a transition from the superhigh-density current pumping to a separate optical pumping scheme

Adiabatic 'transitions' of optical nonlinear materials from bulk to QD materials, and those of signal frequencies from 160 to 640G region

Middle-scale, photonic-logic system integration technology
The highspeed potentials of most of our all-optical gate-circuits are strongly enhanced with using optical interference circuits, up to now. For building, stabilizing, and reducing the foot prints of these partially interferometric circuits, we welcome a merge between the two independent technology, ie. the all-optical circuit technology and the photonic integration technology. It needs long-term research-and-development activities, after preliminary successes by hybrid integration (in NEC and CIP) and monolithic integration (ETH, Alcatel, Princeton, NTT, Univ Tokyo, Alphion, UCSB, UC Davis, etc.) with well-established fabrication techniques and facilities.

Regarding larger-scale photonic-signal-processing systems integration (towards optical computers) with teraherz clocks, we are looking forward to the next innovative advances in the two-dimentional or even three-dimensional photonic-crystal technology.

Research Backgrounds

Why ultrafast, ultrahigh-frequency research?
All the way from the birth of silicon integration technology in 1960's through the on-going nano-technology, what matters has been the length dimension. The number of logic gates or memory cells printed inside one silicon chip in computers and cell phones of today has been scaled up by a factor of 106 or even more from 4 kbits in 1980's through 4 Gbits in 2000's, in inversely proportional to the lithograph-limit length.
In case of magnetic and optical pickups for writing/reading memory disks, what matters has been the length dimension, as well. The number of bytes recorded in one magnetic disk (floppy and hard) and optical disk (MO, 1G-DVD, 2G-DVD) has been scaled up by a factor of about 105, in inversely proportional to the pickup-resolution-limit length, from 128kB to beyond 10 GB nowadays that can store huge amounts of digital databases and many moving videos. In the coming 10-20 years, the mankind will achieve another 102 in the length dimension. In the so-called nano-technology regime originally accelerated by the Clinton administration of the USA, they are literally moving from some tens of nanometers regimes to a few nanometers regimes (10-9 m), in step-wise manners on their long-term roadmap.

Another primary dimension is obviously the time dimension, or equivalently the frequency dimension. In 1960's, semiconductor transistors could not repruduce orchestra sounds in the 10kHz frequency range, as well as the "vacuum tubes" could do at that time. In the first desktop computers in late 1970's, their clock frequency used to be only a few MHz. Now not-special, daily-life computer's clock frequency has reached 3 GHz. The frequency of the fastest transistors of today, which are used in industrial optical-communication systems in trunk networks, have reached 40 GHz.

The question for the coming 10-to-20 years is if and how in principle the mankind will achieve another 102 beyond 40 GHz in the frequency dimension, in energy-effective manners.

Why all-optical gate circuits??
In the end of 20th century, the highest signal frequency (or bitrate) that the fastest electronic transisters can process has reached 40-to-100 GHz. The fastest transistors that have ever been commercialized for civilian-industry applications are those used in 40-Gb/s optical communication systems.
Even between physically 'wired' chips, cards, and boards in those systems, however, the GHz digital signals are no longer 'electronic' signals passing through isolated transistors, resisters, capaciters, inductances, and metalic wires. The GHz signals are behaving as electro-magnatic (em) waves both inside and outside transistor chips, and tend to interfere with each other, radiate to the air, and consequently push up energy dissipation. This radiative behavior of em signal waves is enhanced by the strong velocity dispersion of all metalic, semiconductor, dielectric, and organic materials, since 100-Gb/s digital data signals consist of ultrabroad frequency-spectral components from some hundreds kHz to 100 GHz.

The all-optical gate circuits (ie., optical signal processors) is a distinctively alternative option, where an optical signal processes the other optical signal via an all-optical logic circuit. Its inherent advantage comes from the fact that the carrier frequency of those optical digital signals jumps up to 200 THz (= 2×1012 Hz, in case of standard 1.5-um-wavelength signals). Since even several-Tb/s digital-data bandwidth is only several percentage of the 200-THz carrier frequency, the dispersion-induced radiative behaviors can be more easily suppressed between 'wired' circuit chips, cards, and boards.
On the other hand, the previous disadvantage of all-optical gate circuits was the relatively weak interaction between optical signals in so-called nonlinear optical materials. This disadvantage had caused either very long interaction length (ie., circuit size) or very large optical-energy consumption. Since several technical proposals and inventions in early 1990's, however, their optical energy consumption has dramatically dropped to the order of 100 fJ/bit. The minimum circuit size of the order of 300 by 1,000 um2 is required for such low-energy circuit operation nowadays.
Due to these advantage and disadvantage, we believe that all-optical gate circuits in their very first generation will contribute to ultimately fast systems such as photonic-network-node systems and grid-computer systems.

How on earth do you generate and measure such ultrafast, ultrahigh-frequency optical signals?
For example, it is extremetly difficult to generate 2-ps, 160-GHz electronic data pulses and measure their waveforms and frequency spectra.
It is very difficult, too, to generate 2-ps, 160-GHz optical data pulses and measure them with electronic circuits and conventional EO and OE converters because of their limited speeds.

For the above reasons, the methods to generate ultrafast optical data pulses and measure their waveforms are already, 'all-optical' methods.
When generating 2-ps, 160-GHz optical data pulses, (1) divided-frequency (10 or 40 GHz), 2-ps-short optical clock pulses are generated in semiconductors (MLLD or DISC-loop-type pulse generator), (2) optical clock pulses are digitally encoded into optical data pulses in an OE converter with electronic data signal, and (3) four independent, tributary 40-Gb/s optical data pulses are all-optically time-division-multiplexed to 160-Gb/s optical data pulses.
When digitally receiving the 2-ps, 160-Gb/s optical data pulses, (1) they are all-optically gated to (ie., 4:1 demultiplexed to) 2-ps, 40-Gb/s tributary data pulses, and (2) the 2-ps, 40-Gb/s data pulses are OE-converted back to the original 40-Gb/s electronic data signal.
When measuring their ultrafast optical waveforms, we use all-optical mearement techniques such as cross-correlation or optical sampling, both of which easily provide us time resolution of less than 500 fs or even 200 fs today, instead of electronic sampling scopes whose resolution is limited to 10 ps.

The ultrafast performance of these optical pulse generation and measurement methods are stemming from the fact that an optical short pulse compresses its own width by themselves in an optically nonlinear material to below 100 fs, based on several types of nonlinear-optics principles.
(In good contrast, a microwave pulse never compresses its own width in any material. Nobody has ever found such a material or nonlinear physics.)

These are original, innovative research activities?
After the symmetric-Mach-Zehnder gate structure was proposed to academic societies by Dr. Kazuhito Tajima in 1993, the DISC gate structure was originally proposed by Yoshiyasu Ueno and his co-workers in NEC Corp. in 1998, and patented by the patent offices in Japan and USA. The DISC-loop-type pulse generator was originally proposed by Yoshiyasu Ueno and his co-workers in 2000.

At present, the on-going frequency-scalability viewpoints, quantum-efficiency viewpoints, and some other fundamental-research viewpoints on optical output waveforms and spectra from DISC gates and DISC-loop pulse generator would have been globally recognized as original research activities of this university research group.

In addition, the set of modeling research results on SMZ-3R and PDSMZ-3R gates in 2002-2005 are the other pioneering results from our university group.

Why optical-logic circuit research from now on, in addition to the ultrafast, nano-materials research such as quantum dots?
Before mid 1990's, optical-circuit research activities did not exist; leading-edge optical research activities at those times were categorized into two or three levels, that is, materials (like chemical silicons and metals in electronics), components (like transistors, capacitors, and resistors), and systems (like televisions and computers). Neither of distributed-feedback (DFB) lasers, vertical-cavity surface emitting (VCSEL) lasers, integrated arrayed-waveguide-gratings, nor optical tranceivers had reached optical-circuit levels yet, with respect to microwave em circuits and designs.

Regarding the all-optical gate-circuit research in the last decade since early 1990's, we will recognize them as the first preliminary successes of optical-circuit-level reseasrch and designs in the 100-to-600-GHz region. Well-designed optical circuits (ie., SMZ and DISC) are making the best uses of the ultrafast rise times of less than 100 fs in conventional semiconductor materials (three-dimensional-bulk InGaAs on InP lattice), while the optical circuits are compensating for their drawbacks, that is, relatively slow fall times. We are on our way, as well, to design circuits to compensate for the other second-order waveform distortions as are mentioned above. These research activities to date remind us that an electronic transistor does not by itself amplify an electronic signal nor work as an oscillator. We need to design a circuit for making acceptable amplifiers and oscillators.

Attractive progresses in terms of ultrafast response times or ultrashort-distance interaction lengths, in the meantime, have been made from optical materials research activities such as quantum dots, inter-subband-transition materials, and photonic crystals. In the 21st century, however, it seems to us very difficult to open a new system's world by optical-materials reseach only. Optical-circuit-level research and designs should keep playing their roles; making the best use of the new materials advantages, while compensating for their drawbacks such as second- and third-order-relaxation responses. When expanding this optical-circuit research in the ultrafast regime being supported by more advanced integration technology in the near future, all-optical ultrafast actions of more topological circuits that nobody ever touched will be an interesting new world, too.

What advantage does the "wired" optical communication own, in contrast to "wireless" communication technology?
Until mid 1980's, both nation-wide and world-wide communication systems were relying on wireless (either microwave-repeaters or satellite) communication systems. Since then, most of those wireless microwave systems were replaced with wired optical systems. The advantage of optics comes from its carrier frequency of some hundreds THz, in comparison to the frequency of some tens of GHz of practical microwaves, according to the conventioinal Nyquist criterion.

Regarding the connection bitrate from majority internet subscribers, it used to be only 30-60 kbps in early 1990's. Now we are seeing a dramatic transition from a few MHz with DSL connection, to a few GHz with cost-effective fiber-to-the-home (FTTH) connection.

To summarize, wired optics contributes to bandwidths over Gb/s and over 1 km, while wireless microwave contributes to ubiquitous below Gb/s and 1 km. (The wired optics are thus contributing to the expanding internet infrastructures.)

What is the main issue in the latest, ultrahigh-capacity wavelength-division-multiplexing (WDM) technology?
The latest research-level WDM technology has demonstrated 10-Tb/s digital communication per only one 125-um-diameter silica-fiber cable, over more-than-100-km distance. More practical, WDM systems in the market are already handling nearly 1-Tb/s bandwidths per one cable.

These systems require, however, more than 20-different, 0.8-nm-wavelength-spaced 1550-nm-wavelength components from industrial production lines. Moreover, it is technically difficult to cut down the latency time in cross-connecting data packets between those 20 WDM channels. The reason of these two burdens in the production and architecture is because the digital time-division-multiplexing (TDM) bitrate within these 20 conventional WDM channels is 40 Gb/s.

Why 40-Gb/s TDM system is not enough for us??
なぜ40Gb/s TDMシステムでは不充分なのか?
One 40-Gb/s TDM packet could carry only 400 of 100-Mb/s TDM packets from current FTTH subscribers. It will contain only 40 of 1-Gb/s packets from near-future FTTH subscribers.

In the meantime, the silica-transparecy bandwidth of one fiber cable has reached Δ50 THz (from 1.30 through 1.65 um), while the presently fiber-amp-limited bandwidth is Δ4 THz (in 1.530-1.565 um). The question for the future of mankind is if 1,000 of 40-Gb/s WDM channels is an energy-effective, cost-effective solution or not.

40-THz bandwidth of standard silica fibers (left),
and the 10-Tb/s bandwidth from 273 WDM channels (NEC, 2001, right)

Then, do we develop 40,000 of 1-Gb/s WDM channels per one fiber cable for our future?
では将来、1Gb/s TDMシステムで40,000 WDMチャネルのシステムを開発するか?
The above question will ultimately sound close to another question if we develop 40,000 of 1-Gb/s WDM channels that might directly fit each of our near-future FTTH subscribers. As a matter of fact, nobody will think of this kind of future technology, based on the advantages versus disadvantages from our on-going opto-electronic technology....

Where who is studying 'optical computers," by the way??
Simply speaking, an optical computer where all electronic signals and circuits in laptop computers are replaced with optical signals and circuits requires both (1) speed of all-optical logic circuits, and (2) density of those circuits comparable to Pentium. For us in the first decade in the 21st century, it is a good future target.

It is we who are persuing the (1) direction. So-called photonic-crystal research is needed in the (2) direction.
[Quantum computing, specifically for limited number of categories of mathematical problems, will be between (1) and (2).]

The communication technology in general, for me as a 21st-century citizen
Because of the free internet, many of us in the 21st century can enjoy talking with friends no matter how far they are living, almost free of charge! We can keep in touch with old colleagues, school mates, and sisters/brothers, with living voice (and living video in the near future) as often and long as we like, no matter where we or they actively move in the world. This communication cost looks almost nothing, at present (except cell phones). All of these benefits are coming from the latest optical-communication network infrastructure that started covering our entire globe since early 1980's after wireless, microwave infrastructure in the 1970's. Since early 80's through the recent years, the optical infrastructure has been upgraded repeatedly, in proportional to the growth rate in the computer, the operating system, and the internet.

Rather than voice telephones, you are watching tons of emails, the latest information from public websites, online business transactions or private shoppings, or online software/music downloads? The direct communication between real people is the real communication. In my personal opinion, this communication infrastructure of ours means much more than them in the mankind history; It is the best peace-making infrastructure. It is the only alternative of the transportation infrastructure, which is much better but inherently energy- and time-consuming. Up to now we might have made only 5% wise use of the global inter-networks. Since Johann Gutenberg's invention in the 1430's, it took five centuries for the mankind culture to make the best use of the print-and-publication communication tool!

Now, all of us are receiving the equal benefits from the existing internet? The answer is no. Only metropolitan residents can get benefits of cheap cable-television, DSL, and FTTH connections. Young people born in rural towns or developing countries have to take care of expensive living costs, for their going to attractive universities and industry in big cities. When another billions of people start using broadband accesses (including e-campus accesses, for instance) to the free internet, will it sustain the total bandwidths? When the present expensive costs of cell phones suddenly drop by a factor of ten, the presently free IP-phone networks will sustain the total bandwidths? With the existing optical network technology, the answer will be no. The existing optical technology will not be fast enough, will not be cheap enough. The existing broadband infrastructure seems to have started consuming much more energy than that in 1990's, too. Nowadays, broadband network computers (10G/port) can melt down, without jet-engine-like cooling fans!

We need better optical communication systems. We wish several more global groups join our future research fields.
Let's open an innovative new door to the future!

University of Electro-Communications, Graduate School of Electronic Engineering / Dept. of Electron. Eng.
Ultrafast Optical Logic Laboratory
Yoshiyasu Ueno, associate professor