As a result of the growth of broadband communications through the spread of asymmetric digital subscriber line (ADSL) and fiber-to-the-home (FTTH), we can expect to see a rapid increase in the communication capacities of backbone optical networks. To keep pace with such a rapid growth of data-com traffic, the number of wavelength channels in dense wavelength division multiplexing (DWDM) transmission system will evolve continuously. Upgrading a basic bit rate in a commercial system from the current 10 Gb/s to a higher value is a promising way to reduce the managing efforts of ever-expanding optical networks. Although, today, 40-Gb/s transmission technology has entered the stage of commercial application, it would be necessary to extend transmission rate to 160 Gb/s or higher for the future ultrahigh-capacity photonic network. As the maximum speed of available electrical-time-division-multiplexing (ETDM) circuit is currently 100 Gb/s [1]– [2]
[3], the so-called optical-time-division-multiplexing (OTDM) technique has been widely used to increase transmission rate up to 160 Gb/s. The OTDM format enables us to increase channel rate without electrical limitation, and so far, terabit-capacity OTDM transmission has been demonstrated with single wavelength channel [4], [5]. 160-Gb/s-based OTDM/WDM hybrid transmission experiments have also been reported, and the feasibility to reduce wavelength channels in terabit-capacity WDM transmission has been proved [6]– [7]
[8]. However, the management of optical signal-to-noise ratio (OSNR), dispersion compensation, fiber nonlinearity, and polarization mode dispersion (PMD) tend to become more and more strict in the 160-Gb/s-based transmission systems. To ease such severity of system management, recently, forward-error-correction (FEC) [8] and a novel phase-coded signal represented by carrier-suppressed return-to-zero (CS-RZ) [9], [10] or RZ-differential-phase-shift-keying (RZ-DPSK) [11] have been introduced to 160-Gb/s OTDM signal, and the improvement of transmission capability was demonstrated experimentally. Also, against the issue of severe dispersion tolerance, an adaptive dispersion compensation technique, which is able to reduce the influence of local dispersion fluctuation induced by variation of environmental condition, has been successfully developed [12]. In addition, the advance of adaptive PMD mitigation technique [13] and optical 3R regenerator [14] strongly supports the potential to realize future 160-Gb/s-based optical links.
However, despite the continuous progress of 160-Gb/s signal transmission technology, an optical multiplexing technique is still in a developmental stage. As the optical multiplexer for the 160-Gb/s signal generation, the passive-delay-line optical circuit has been used widely. Although it is convenient to make a tentative 160-Gb/s signal, the authentic multiplexer, which essentially requires an individual encoding of all multiplexed channels, should be employed from a practical point of view. So far, we have developed the authentic 4
$\,\times\,$40-Gb/s optical multiplexer integrated four electroabsorption (EA) modulators [10]. Since the EA modulator can be driven by very simple principle, it should make the handling of high-speed optical signal easier, and consequently, lead to stabilize OTDM signal generation. Besides our approach, the planner lightwave circuit (PLC)-based all-optical multiplexer integrated periodically poled lithium niobate (PPLN) has been reported, and all-channel-independent encoding has been achieved successfully [15].
In this paper, we describe a simple optical time-division multiplexer in which EA modulators are installed for true 160 Gb/s signal generation. A unique feature of the optical multiplexer is that it can introduce advanced modulation format, for example CS-RZ, to 160-Gb/s OTDM signal [10]. We show some examples of OTDM signal generation with various modulation formats. We also describe the prototype of a 160-Gb/s OTDM transmitter employing the optical multiplexer and a receiver based on EA modulators. Its practicability is discussed through experimental results of 160-Gb/s long-distance field transmission on the Japan Gigabit Network II (JGN II) optical testbed.
SECTION II
160-GB/
$ {\rm s}$ OPTICAL TIME-DIVISION MULTIPLEXER
An optical multiplexer, which converts optical pulse train to 160-Gb/s OTDM signal, performs data modulation for each multiplexed channel individually, thus requiring multiple optical modulators. The optical multiplexer with a basic rate of 40 Gb/s requires four optical modulators. It is important that these modulators are installed efficiently within a compact package. Accordingly, we have developed an optical multiplexer based on a free-space integration technology. Fig. 1(a) and (b), respectively, illustrate a schematic diagram and a general view of the 4
$\,\times\,$40-Gb/s optical multiplexer. The optical modulators adopted here are EA modulators, which are small and offer high-speed modulation with a bandwidth of 35 GHz. The four EA modulators are individually encapsulated in hermetically sealed packages to secure stability and reliability in modulation and are placed in four deferent spatial paths composed of half mirrors and total-reflection mirrors. Each of the four 40-Gb/s electrical data signals is directly supplied to the respective modulator by connecting a co-axial cable with V-band connector. Since there does not exist any electromagnetic crosstalk between electrical data signals, it greatly reduces the complexity in a high-speed electrical design. Each of the modulators converts an input 40-GHz optical pulse-trains into 40-Gb/s optical signals. Each of the four 40-Gb/s optical signals is sequentially bit-interleaved to 80 Gb/s, and then, to 160 Gb/s, and two sets of 80-Gb/s signals and 160-Gb/s signals are then output. Two 80-Gb/s outputs and a 160-Gb/s output (80-Gb/s OUT 1, 2, 160-Gb/s OUT 2) are used as reference signals for alignment of modulation timing and for carrier-phase monitoring. The insertion loss of the optical multiplexer is 15 dB and the loss deviation between multiplexed channels is less than 0.5 dB. The delay-time accuracy of bit-interleaving was estimated to be better than 10 fs (6.28
$\ \pm\ $0.01 ps).
The developed optical multiplexer is unique in that the optical carrier-phases of all multiplexed channels are controlled separately. The optical carrier-phase difference between adjacent bits is an important factor for fiber transmission, since nonlinear pulse-to-pulse interaction depends on the phase correlation. However, the carrier-phase difference, particularly in fiber-based optical multiplexer, is usually indefinite and unstable, since the multiplexed channels experience different effective optical lengths, which are much longer than the signal wavelength and are sensitive to the variation of environmental condition. Therefore, the transmission performance would become unstable. In the optical multiplexer shown in Fig. 1, undesirable variation of the effective optical length for each multiplexed channel is small enough, thanks to the free-space design. Thus, the carrier-phase relation between the multiplexed channels is stably kept to be constant. The stability of carrier-phase was estimated at 0.08 rad/°C. The more beneficial feature of the optical multiplexer is that the relative carrier-phases of multiplexed channels can be controlled optionally with changing operation temperatures of EA modulators. As the temperature change brings about the deviation of effective refractive index in an EA waveguide by 5
$\,\times\,{\hbox{10}}^{-4}[{\hbox {1}}/^{\circ}{\rm C}]$, a phase-shift of
$\pi$ rad should be achieved with the temperature change of 8 °C under the assumption of 200-mm-long EA waveguide. Fig. 2 represents the output power variation of 80-Gb/s OUT 2 as a function of the operation temperatures of EAM 2 and EAM 4. The output power was measured by inputting a continuous-wave (CW) light instead of a pulsed signal. A signal wavelength was set at 1548 nm, and a bias voltage applied to each EA modulator was set at 0 V. The output power varied sinusoidally depending on the temperature of EAM 4, and the condition minimizing interferometer output moved with increasing the temperature of EAM 2. It indicated that the relative carrier-phase between multiplexed signals is to be adjusted freely by the temperature change. Insertion loss change of EA modulators is less than 1 dB under the operation temperature ranging from 15°C to 35°C. Thanks to this feature, the optical multiplexer enables the generation of phase-coded signal such as CS-RZ signal even at a bit-rate of 160 Gb/s. The CS-RZ modulation format is well-known format tolerable to nonlinear pulse-to-pulse interactions [16] and would contribute to improvement of 160-Gb/s transmission performance [10]. In order to generate 160-Gb/s CS-RZ signal, a continuous phase state is imposed on each of the two 80-Gb/s signals prior to bit-interleaving. The continuous-phase 80-Gb/s signal can be obtained easily by adjusting temperatures driving two EAMs, as shown in Fig. 2. The relative carrier-phase between two 80-Gb/s signals is to be aligned by adding an offset-temperature to the condition giving continuous carrier-phase. Fig. 3 illustrates CW output power variation when the offset temperature was added to operation temperatures of EAM 2 and EAM 4. The temperatures driving EAM 2 and EAM 4 were initially tuned at 25°C and 28.5°C, and were then, given the offset temperature while maintaining the continuous-phase condition. Here, the temperatures of EAM 1 and EAM 3 were fixed to be 21.5°C and 21°C, respectively. As the relative carrier-phase between two multiplexed beams should vary depending on the offset temperature
$\Delta T$, the output power at 160-Gb/s ports sinusoidally changes, as shown in Fig. 3. In accordance with the nature of the interferometer, the output powers launched from two 160-Gb/s output ports are complementary each other, so that a continuous-phase signal and
$\pi$ radians out-of-phase signal should be launched simultaneously when a pulsed-signal is input.
For precise control of carrier-phase, it is important to monitor the carrier-phase relation with good accuracy. The optical phase state of each output can be estimated by free-space 1-bit delay interferometer, which is widely used as optical demodulator of DPSK signal. Since the output power of the interferometer strongly depends on the carrier-phase difference between adjacent bits, the phase information is transformed into a function of optical power. By use of the phase monitoring method, we have also developed an adaptive phase controller that is described elsewhere [17].
SECTION III
160-GB/
$ {\rm s}$ OTDM TRANSMITTER AND RECEIVER
A. 160-Gb/s Otdm Transmitter
Fig. 4 shows the block diagram of a 160-Gb/s OTDM transmitter, consisting of an optical pulse source and an optical multiplexer. The optical pulse source generates a 40-GHz pulse train, and it is launched to the 160-Gb/s optical multiplexer. At the multiplexer, the pulse stream is converted to four 40-Gb/s data streams, and is then, bit-interleaved to 160-Gb/s signal. In such an OTDM scheme, it is necessary to generate optical short pulse train less than 3 ps with an extinction ratio of more than 35 dB in order to prevent intersymbol-interference (ISI). In addition to such requirements, the optical pulse source has to be highly stable and compact in terms of practicability. Semiconductor-based mode-locked laser diode (ML-LD) [18], [19] is known as a suitable pulse source to meet such requirements, and the applicability to OTDM signal has been investigated widely [20]– [21]
[22]. As a simpler alternative, an external modulation scheme utilizing high-speed EA modulator was also employed widely [23]– [24]
[25]. The optical pulse generation with the external modulation supports arbitrary clock frequencies in principle and can generate extremely stable optical pulses using low-jitter modulation signals. Here, the system uses two EA modulators connected in series to generate optical pulse train that can be applied to the 160-Gb/s OTDM signal [26], [27]. The two EA modulators are driven by the 40 (39.81312)-GHz clock with adjusted timing and convert the CW beam into Gaussian pulses. A bias voltage applied to each EA modulator was
$-3$ V, and an RF power of modulation signal was 20 dBm. The average light powers launched at both EA modulators were set at 10 dBm. Fig. 5 represents the autocorrelation trace of output pulse train at signal wavelength of 1548 nm. The pulse width was about 2.8 ps and the extinction ratio was expected to be about 35 dB. The pulse shape was close to Gaussian and the time bandwidth product was estimated at 0.46, which indicated output pulse was nearly Fourier-transform limited pulse. The EA-modulator-based pulse source is also applicable in the wavelength range from 1540 to 1560 nm, changing the driving condition slightly. According to the general characteristics of EA modulator, an optical loss increases in the shorter wavelength region. Although it led to an increase in amplified spontaneous emission (ASE) noise in signal, an OSNR greater than 28 dB (resolution bandwidth 1 nm) was maintained even at the wavelength of 1540 nm. The OSNR value is high enough to achieve theoretical
$Q$-factor greater than 30 dB in the 160-Gb/s OTDM signal.
At the optical multiplexer, 40-GHz pulse train is converted to four streams of 40-Gb/s optical data, and then, bit-interleaved to 160 (159.2525)-Gb/s OTDM signals. As described in Section II, the optical multiplexer enables the generation of the phase-coded OTDM signals. Optical spectra and optical sampling traces shown in Fig. 6 demonstrate simultaneous generation of CS-RZ signal (160-Gb/s OUT 1) and continuous-phase RZ signal (160-Gb/s OUT 2). The operation temperatures of four EAMs were adjusted to 20°C, 31 °C, 22°C, and 35°C, respectively. The 40-Gb/s ETDM signals with the pseudo-random-bit-sequence (PRBS) of
$2^{15} -1$ were used to modulate four EA modulators. The electric power of 40-Gb/s data signal was 14 dBm, and the bias voltage applied to EA modulator was, typically,
$-1.5$ V, which changes slightly depending on the operation temperature. The CS-RZ coding is dealt with in such a way that the optical pulses acquire the phase alternating between 0 and
$\pi$ at every time slot, so that the carrier component in the CS-RZ signal spectrum is to be canceled out. The optical spectra exhibit the spectral shapes peculiar to their respective modulation formats, and there is no clear evidence of modulation-data correlation, thanks to the independent encoding. Although the slight nonuniformity of pulse-peak power was observed, particularly, in the conventional RZ signal, the eye diagrams showed clear eye-opening. Recent theoretical and experimental works have predicted that effective suppression of intrachannel four-wave-mixing [28], which is known as one of the limiting factors in fiber transmission, is achieved with new modulation formats. The modulation scheme is characterized by an alternating phase between 0 and
$\pi /2$, i.e., “
$0\,\pi /2\,0\,\pi /2$” (
$\pi /2$-APRZ), or pair-wise alternating phase between 0 and
$\pi$, i.e., “
$0\,0\,\pi \,\pi $” (PAP-CSRZ) [29], [30], and is also realizable utilizing the carrier-phase tunability of the developed optical multiplexer. Fig. 7 represents optical sampling waveforms and optical spectra of such phase-coded signals. The
$\pi /2$-APRZ signal having a phase code of “
$0\,\pi /2\,0\,\pi /2$” corresponds to a transient state transforming continuous-phase signal into
$\pi$-radian out-of-phase signal. So, both carrier component and sideband components 80-GHz apart are observed in the optical spectrum. The pulse tail overlapping in the eye-diagram is larger than CS-RZ, but smaller than continuous-phase RZ. In the PAP-CSRZ signal phase-coded with “
$0\,0\,\pi \,\pi$,” the relative phase alternates between 0 and
$\pi $ with the period of 80 GHz, so that the spectrum is equivalent to 80-Gb/s CS-RZ. As shown in the eye-diagram, large pulse overlapping like continuous-phase RZ and no overlapping like CS-RZ alternates. The 160-Gb/s signal with
$\pi /2$-APRZ can be generated only by setting the relative phase difference between two continuous-phase 80-Gb/s signals at 0 or the multiple of
$2\pi$, prior to bit-interleaving at 160 Gb/s. Regarding the case of PAP-CSRZ, the 80-Gb/s signals to be bit-interleaved must be CS-RZ with the relative phase difference of 0 or the multiple of
$2\pi$. In any case, two 160-Gb/s signal with the same modulation format should be launched from the multiplexer, unlike the case of CS-RZ generation.
B. 160-Gb/s Otdm Receiver
Fig. 8(a) represents the configuration of 160-Gb/s receiver with an EA-modulator-based optical demultiplexer and a clock recovery circuit. Incoming 160-Gb/s signal is optically demultiplexed to 40 Gb/s by the EA modulator driven by a 40-GHz clock, and is then, sent to a 40-Gb/s receiver. Strictly speaking, four EA modulators are required for a 1:4 optical demultiplexing as well as the multiplexer. Also, in this case, the authentic demultiplexing module will be realizable by employing the free-space structure similar to the optical multiplexer. Fig. 9 corresponds to the optical sampling traces of optically demultiplexed 40-Gb/s signal. The gate width of the demultiplexer was measured at 4 ps by inputting CW light instead of 160-Gb/s signal. The sampling trace shows that the 40-Gb/s signal was clearly separated from the 160-Gb/s signal. The suppression ratio of neighboring channels was estimated to be greater than 15 dB. Since the polarization state of the incoming signal is not always constant, the polarization insensitivity is an essential requirement for the stable operation of optical demultiplexer. The polarization-dependent-loss (PDL) of the EA modulator is designed to be less than 0.5 dB by applying tensile strain to InGaAsP-based multiple-quantum-well (MQW) EA waveguides, so that the polarization-insensitive demultiplexing is secured. The 40-GHz clock signal is recovered with an opto-electronic hybrid phase-locked loop (H-PLL) shown in Fig. 8(b) [31]. A polarization-insensitive EA modulator is installed at front-end of PLL circuit and is driven by a locally generated clock signal, which is derived from a voltage-controlled oscillator (VCO). As a small frequency offset
$\Delta f$ (250 MHz) is added to the clock signal, the output optical signal from the EA modulator includes the beat frequency component of
$4\Delta f$ (1 GHz). The optical output beating at
$4\Delta f$ is converted to an electrical signal using O/E converter (3 dB bandwidth
$= 1.7$ GHz) and bandpass filter (
${\hbox {BPF}}, {\hbox {center frequency}} = 1$ GHz). Then, the phase deviation between the prescaled signal and the reference signal
$(4\Delta f)$ coming from a local oscillator (LO) is detected at a phase comparator to facilitate PLL operation. The root-mean-square (rms) timing jitter of the recovered 40-GHz clock was, approximately, 60 fs. The capture range and the locking range of PLL operation were measured at 1.2 and 12 MHz, respectively. The robustness of PLL operation was evaluated concerning 160-Gb/s signals distorted by chromatic dispersion or PMD, and it was confirmed that the locking characteristic was, still, stable within a range of chromatic dispersion between
$-5$ ps/nm and 5 ps/nm and within a range of differential group delay (DGD) by PMD between 0 and 3 ps.
C. Back-To-Back Performance
Fig. 10 represents the back-to-back performance of the developed equipment. Bit-error-rate (BER) was measured for each of the four 40-Gb/s demultiplexed signals, varying a decision level in bit-error evaluation. At the transmitter, the PRBS of four 40-Gb/s ETDM signals was set at
$2^{15} -1$. A 160-Gb/s signal was tuned to be CS-RZ signal. Fig. 10 corresponds to the results of a 40-Gb/s tributary that exhibited the worst BER performance. The measurement was performed continuously for 300 min, and the
$Q$-factor [32] estimated from the BER performance was 27 dB on average. This value corresponds to a BER value of
$10^{-100}$ or less at the optimum decision point, attesting to excellent back-to-back performance. The
$Q$-factor variation during 300-min continuous measurement was less than 1.3 dB, and it revealed an extremely stable performance of the equipment, which is an essential factor in practical applications.
The same evaluation in the signal wavelength of 1540 and 1560 nm also revealed equivalent back-to-back performances. Table I summarizes
$Q$-factors and OSNR, measured at wavelengths of 1540, 1548, and 1560 nm. The OSNR was measured at transmitter end. It was confirmed that the
$Q$-factor greater than 26 dB was obtained in a wavelength range of more than 20 nm. Slightly worse
$Q$-factor in shorter wavelength originates in OSNR degradation due to larger insertion loss in EA modulators. It is thought that the result is what proves the wideband applicability of the EA-modulator-based OTDM system quantitatively. The back-to-back performances of the other three modulation formats described earlier are also summarized in Table II. The
$Q$-factor of each modulation format exhibits very close value, and it suggests that the developed prototype equipment is useful for transmitting ultrahigh-speed optical signal with various modulation format at wide wavelength range.
SECTION IV
160 GB/S FIELD TRIAL
For ultrahigh-speed transmission of up to 160 Gb/s, the degradation in transmission characteristics due to PMD of the optical fiber transmission line represents a serious problem. Among other experiments, a field transmission experiment using installed optical fiber cables has demonstrated the necessity of applying the PMD compensation technique [13], [33]– [34]
[35]. Unlike in relatively stable laboratory environments, for installed optical fibers in practical use, PMD and chromatic dispersion varies according to external factors such as weather conditions [33]. For practical applications, the desired transmission characteristics must be maintained under circumstances in which the optical fiber characteristics are subject to continuous change. In such a situation, the stability and high-performance of a transmission equipment becomes more and more important, especially when the bit-rate is increased. In order to verify the practicability of the developed 160-Gb/s transmission equipment, we performed a field transmission experiment using an optical testbed in JGN II. Fig. 11 shows the network configuration of the optical testbed and the experimental transmission system. The optical testbed consists of ten standard single-mode fibers (ITU-T G.652 SMF) installed between the Keihanna Human Info-Communication Research Center (Seika Town, Kyoto) and the Dojima Base Station (Dojima, Osaka). Each transmission line is 63.5 km long, offering a total link length of 635 km, with ten loop-back paths. Optical amplifiers [erbium-doped fiber amplifiers (EDFAs)] are installed at Keihanna and Dojima to form a ten-span link structure with a repeater spacing of 63.5 km. The transmission loss per span is, approximately, 15 dB. The optical amplifier has a two-stage configuration with a dispersion compensating fiber (DCF) between the stages. The residual dispersion per span is typically
$+10$ ps/nm, and the dispersion slope is nearly 100% compensated. Chromatic dispersion over the entire transmission line is adjusted at 0 ps/nm, with the additional DCFs installed at input-end and output-end of transmission line. To reduce the nonlinear impairment in transmission performance, the CS-RZ modulation is introduced in the transmitted 160-Gb/s signal. Slight precompensation of the dispersion (performed by installing a DCF prior to transmission) also helps reduce signal degradation attributable to fiber nonlinearity [36]. The signal input levels into the SMF and the DCF were set at 4.5 and 1 dBm, respectively. A PMD compensator (PMDC) was placed in front of the receiver. The PMDC features the first-order PMD compensation scheme consisting of a polarization controller, a tunable DGD generator [37], and a polarization analyzer to detect the degree of polarization (DOP). The PMDC compensates for PMD by adjusting the polarization of the input signals and the extent of DGD generation, maximizing the DOP of the output signals. In the experiment, the PMDC was controlled manually.
Fig. 12 shows the results of
$Q$-factor measurement performed every 127 km from 254 to 635 km. The
$Q$-factor was evaluated for all tributary channels demultiplexed to 40 Gb/s from 160 Gb/s. We obtained good transmission characteristics with a
$Q$-factor of 16.2 dB (
${\rm BER} \approx 10^{-10}$) on average over a transmission distance of 508 km, which corresponds to the distance between Tokyo and Osaka. The
$Q$-factor was maintained below 15.7 dB (
${\rm BER} \approx 10^{-9}$) on average even after transmitting over 635 km. Fig. 13 shows the optical sampling waveforms at back-to-back and after transmitting over 635 km. It is confirmed that 160-Gb/s signal waveform still exhibits clear eye-opening even at the distance of 635 km. The
$Q$-factor fluctuation illustrated in Fig. 12 was mainly due to the time-dependent variation of PMD, which changed according to the external environment. Because of manual operation of PMDC, the perfect compensation of PMD was not maintained for the long term, so that the transmission characteristics varied with time even at the same transmission distance. It has been confirmed that the amount of DGD due to PMD changes from 3 to 7 ps at a transmission distance of 635 km, and slight discrepancy from the optimum operation condition of PMDC results in serious degradation of signal quality. Fig. 14 shows the results of continuous
$Q$-factor measurement, performed at transmission distances of 381 and 508 km. The arrows indicate the time at which the PMD compensator was re-adjusted in response to
$Q$-factor degradation. In both cases, the PMD compensator required readjustment after 10–20 min in order to maintain error-free performance. Nevertheless, the results indicate that introducing the automatic PMD compensation technique in conjunction with the developed 160-Gb/s transmission equipment will secure the long-term stability required for practical applications. In terms of the accuracy on DGD cancellation in PMDC, experiments and simulations have indicated that the degradation of signal quality is slight with a DGD of 1 ps. For these reasons, we have concluded that the adaptive PMD compensation technique, which is technically mature as a 40-Gb/s transmission technology, can meet the necessary requirements. In the same vein, although this experiment did not lead to the detection of any distinct effects, it has been reported that the effects of higher order PMD are also serious in 160-Gb/s long-distance transmission [35]. We believe that the design of the transmission system must take these possible effects as well into consideration.
160-Gb/s ultrahigh-speed optical transmission is expected to become an important fundamental technology in the construction of next-generation high-capacity optical networks. Accordingly, we have been developing 160-Gb/s optical time division multiplexing/demultiplexing techniques based on EA modulators, focusing on practicality and operability. In the course of our activities, we have developed a free-space integrated optical multiplexer that supports individual data modulation for each of the multiplexed signals and have established a technique for generating stable and high-quality 160-Gb/s OTDM signals while maintaining practicability. These are regarded as core technologies for generating ultrahigh-speed OTDM exceeding the processing speeds of electronic devices. CS-RZ is an example of a superior modulation format suitable for ultrahigh-speed transmission. We have also enabled the introduction of such formats into 160-Gb/s signals, taking advantage of the stable carrier phase, a characteristic of a space-integrated optical multiplexer. In a field transmission experiment using the JGN II optical testbed connecting the Keihanna Human Info-Communication Research Center (Seika Town, Kyoto) and the Dojima Base Station, we have demonstrated satisfactory 160-Gb/s ultrahigh-speed long-distance transmission over 635 km by use of the developed 160-Gb/s OTDM transmission equipment.
Minor problems remain, such as the long-term stability of transmission performance. However, along with rapidly progressing optical signal processing techniques, involving modulation format, adaptive chromatic dispersion, PMD compensation, and 3R regeneration, we are optimistic that efforts toward practical application of our ultrahigh-speed transmission system will soon yield results.