Open Access
Add to BibSonomyAbstract
In order to monitor quality of ultra high bit-rate optical signals in a future optical network, such as 160 Gb/s, a simple monitoring technique is required. Therefore, a novel waveform monitoring technique by prescaled-clock tone detection was proposed in a previous report. In this paper, detailed principle of the proposed technique was explained. The monitoring technique is based on an asynchronous beat signal generation using an elecro-absorption modulator (EAM) and is able to separately observe waveform distortion caused by accumulated chromatic dispersion (CD), polarization mode dispersion (PMD) and optical signal-to-noise ratio (OSNR) degradation. The verification of concepts was performed by experiments, in which 1 GHz pre-scaled signals were employed to monitor distortion of OTDM 160 Gb/s carrier suppressed return-to-zero (CS-RZ) signals. Furthermore, applicability to Q factor estimation was verified by an experiment. In addition, an observation of 160 Gb/s signal by the proposed monitor was demonstrated over 120 minutes using an installed fiber in JGN-II testbed.
©2009 Optical Society of America
1. Introduction
Rapid advancement of fiber to the home (FTTH) causes increase of densely wavelength division multiplexing (DWDM) channels. In addition, achievement of higher bit rate has been required to restrain the increase of system power consumption owing to reduction of the WDM channels. Therefore, terabit-capacity transmission experiments with single channel optical time division multiplexing (OTDM) system [1]–[3] and 160 Gb/s OTDM/DWDM hybrid systems [4]–[6] have been reported, and they revealed high-potential to realize the future ultra high-capacity optical network.
On the other hand, Management of the signal condition is required to maintain the high service quality. Higher bit-rate signals, such as 160 Gb/s, enable to monitor the signal quality more easily by reduction of WDM channels. In addition, optical signal monitoring technique without optical-to-electrical conversion promotes power consumption by reduction of electrical integrated circuit. So far, several monitoring techniques of the optical signals directly have been reported [7]–[21]. Especially in these monitoring methods, asynchronous techniques are required in respect that they enable to monitor signal with simple equipments. They were roughly classified into time-averaged and time-resolved methods by Kozicki et al. [11].
Time-resolved methods are based on amplitude histogram such as an asynchronous sampling Q monitor [12], histogram Q monitor [13] and delay-tap sampling monitor [14], [15]. These techniques are suitable for intuitive understanding because of their feature to display waveforms. Even in achieving 160 Gb/s of signal bit rate, delay-tap sampling techniques have possibility to be utilized by adjusting the delay time between sampling pair. Though time-resolved monitoring techniques have such advantages, they need inevitability two-dimensional analysis, which make systems comparatively complex.
Time-averaged monitoring techniques output not waveforms but some physical values, e.g. intensity [16]–[21]. Therefore, complex postpositional analysis of the detected signal is not necessary for them. Furthermore, unnecessity of modification to transmitter is more preferable to keep systems simple. As achievement of such simple techniques, signal distortion by chromatic dispersion (CD) and differential group delay (DGD) monitors using clock tones [18]–[21] and peak power by two photon absorption [21] have been reported. One of problems of utilizing clock-tone measurement to 160 Gb/s signal is bandwidth limitation of electric circuit. To overcome this difficulty, we have developed novel waveform distortion monitoring technique for 160 Gb/s signals by observing pre-scaled 1 GHz signals and the concept of the monitoring technique have been proposed [22]. The proposed waveform distortion monitor is based on pre-scaled asynchronous signal detection which generated by a sinusoidallly driven EA modulator.
In this paper, we describe detailed principle of the pre-scaled technique in a general case. The clock tone of input signal with frequency fs Hz is converted to nΔf Hz prescaled signal by the proposed EA prescaler, which is sinusoidally modulated with frequency of fs/n-Δf Hz. The prescaled signal intensities are in proportional to the clock tone and the frequency of nΔf is very low compared with fs. Therefore, the prescaled beat signals are useful to decrease the passband of the detection circuit and especially to measure the clock tone intensity that exceeds the passband limit of a detection circuitt. In the opening of section 2, generation of the prescated signal is explained and limits of fs and nΔf are discussed. In the next place, we show application of this technique to monitor optical signal-to-noise ratio (OSNR) degradation, CD and DGD by calculated results. In the last of this section, we explain the techniques to distinguish the origins of distortion from the beat signals. To confirm the principle, we demonstrated the experiments using carrier suppressed return–to-zero (CS-RZ) 160 Gb/s signals [23] and the result is shown in section 3. Furthermore, we show an observation of 160 Gb/s signal after 127 km field transmission over 120 minutes by the proposed monitor and an installed fiber in Japan Gigabit Network (JGN)-II testbed [24]. In section 4, we demonstrated applicability of the technique to signal performance monitoring, such as Q factor estimation, by detecting the input-to-output ratio of the prescalar.
2. Principle
2.1 Generation of prescaled signal
In this section, generation of prescaled signal is explained in a general case. Fig. 1(a) shows the schematic diagram of the proposed prescalar detecting fs Hz clock tone using an EA modulator with bandwidth of fs/n Hz. The frequency of input signal is fs Hz and the driving signal frequency fm is fs/n Hz with offset frequency of Δf Hz. Fig.1 (b) illustrated schematic drawing of the principle. An fs Hz pulse train, which is shown by dotted line, is input into the EA modulator. The input pulse train is gated by the sinusoidal modulation with fs/n−Δf Hz at the EA modulator. In this figure, the peak of the first gate pulse (the order i is 0) is assumed to correspond to the peak of input pulse. The peak of the second gate pulse (i=1) have time difference of t0 against the peak of input pulse with the order j of n, and the time difference of t0 is given by
The peak of gate pulse with the order i of m overlapped the input pulse peak with the order j of nm+1. Consequently, frequency of beat signal is given by
where, mt0 corresponds with 1/fs. Therefore, nΔf Hz signal is generated by the fs/n-Δf Hz sinusoidal modulation. The nΔf Hz signal is detected by a radio frequency power meter (RF-PM) through an optical-to-electrical (O/E) converter and a bandpass filter (BPF) as shown in Fig. 1(a). The intensity of the nΔf Hz generated signal is discussed in the next place.
For simple discussion, Transmittance of the EA modulator and input signal are set as
where h is insertion loss of EA modulators and k is a coefficient of increasing loss with applied voltage. V(t) is an electrical driving signal with amplitude of V0 and frequency of fm and set as
Output signal is given by product of Eq. (3) and Eq. (4). Eq. (5) is substituted to Taylor series of Eq. (3). Consequently, nΔf Hz beat signal intensity is given by
where I1, I2 and I4 are intensities of beat signals in case that n equals to 1, 2 and 4, respectively. Calculated intensities by Eqs. (6), (7) and (8) are illustrated in Fig. 1(c) by lines. Parameters of h and k are 0.25 and 0.6, reflected an actual EA modulator characteristic. In this figure, simulated results were plotted by marks with the calculated results. The simulated results correspond with calculated ones up to about 2.5 V of applied voltage. The intensities of beat signals with the multiple integers of up to 8 are represented in Fig. 1(c) and they have enough intensity to observe clock tones in the optical signal instead of the immediately measuring. Beat signal with the multiple integer n of 16 is too small to detect the clock tone.
Assuming the passband limit of a practical electric circuit is 50 GHz, the proposed technique is able to be utilized up to 400 Gb/s signal monitoring. Since the beat signals were generated corresponding to the shape of envelope curves, the proposed technique is able to apply to any shift keying, such as differential phase shift keying (DPSK) and quadrature phase shift keying (QPSK), if they have return-to-zero (RZ) signals. Clock tones of return-to zero (NRZ) signals, which are generated by signal distortion [18], are less than spectral intensity by data coding. Therefore, the proposed technique is not suitable for monitoring deterioration of NRZ signals.
Fig. 1. Principle of proposed waveform distortion monitor, (a) prescalar using EA modulator, (b) schematic drawing of beat signal generation in the EA modulator, (c) intensity of generated beat signals, calculated by Eqs. (4)–(6) (lines) and numerical simulation results (marks) in case n equals to 1 to 8 (simulated results only in case n=8), (d) practical setup of the waveform distortion monitor.
For practical use, the clock tone is evaluated compared with the total signal power. The prescaled monitor is equipped with a prepositional optical bandpass filter (OBPF) and a power meter to detect total power. The set-up is illustrated in Fig. 1 (d). Output of the EA prescalar is converted to an electrical signal with a photo diode (PD) and amplified to detect a radio frequency power meter (RF PM). The averaged total power was monitored by detecting tapped optical signal with a PD and a low pass filter and a power meter (PM). The ratio of the nΔf Hz beat signal and the input signal (a/b) was defined as input-to-output intensity ratio (I/O IR) in the following.
The electrical BPF that is inserted before the RF-PM eliminates unnecessary frequency components of the beat signal. The filter characteristic is designed in consideration of n, Δf and stability of the oscillator (δf). That is, passband of a filter (+/−f1) was affected by δf and sideband suppression ratio (SSR) was decided to eliminate neighboring beat frequencies (+/−nΔf). Providing flatness through passband of the filter, it is assumed that the filter has Butterworth characteristic. Order of Butterworth bandpass filter is given by Eq. (9) in case with passband +/− f1, flatness F, and SSR in frequency of +/−f2 [25]
Figure 2 is calculated result of Eq. (9) as the function of nΔf (equal to f2/2). In these calculations, f1 is assumed 20 MHz in reflection of stability of an actual oscillator. Considering practical loss of filters is less than 3 dB, it is desirable that the order of filter is 8 or less. This condition requires that nΔf is higher than about 250 MHz from the result shown in Fig. 2. In the next place, we discuss I/O IR decrease by signal distortion.
Fig. 2. Product of n and Δf vs. order of filter, a minimum integer more than the calculated value is used as an order of filter response design. F is flatness and SSR is sideband suppression ratio, and stability of oscillator is assumed +/-20 MHz.
2.2 OSNR degradation
When signals are distorted by optical signal-to-noise Ratio (OSNR) degradation, the clock-tone decreases compared to the input signal, and consequently I/O IR of the monitor is reduced. The relation of the I/O IR and OSNR measured by a spectrum analyzer is given by
where Δλ is bandwidth of a prepositional OBPF, and res is resolution bandwidth (RBW) of the optical spectrum analyzer (OSA). Calculated results by Eq. (10) are illustrated in Fig. 3. In this figure, res is 1 nm and Δλ is 4, 10 and 20 nm. The I/O IR decreases according to the OSNR degradation. Here the I/O IR of 0 dB set at OSNR of 30 dB. It is expected that OSNR degradation is measurable with the I/O IR from results shown in Fig. 3.
2.3 Chromatic dispersion
Waveform distortion by accumulated chromatic dispersion (CD) causes the signal pulses widening and the decrease of fs Hz intensity. The signal pulses widening by CD has very close relations in the initial pulse width. Therefore, special case of 160 GHz as fs is discussed in the following. The clock-tone in the 160 GHz short pulse train is expressed in first term of Fourier series. Let assume chirpless Gaussian as the signal pulses, and accumulated dispersion and 1/e pulse width be Da and a0. The ratio of total signal intensity and 160 GHz component is then given by
where, λ and c are signal wavelength and speed of light in vacuum, respectively [26].
Fig. 4 shows I/O IR vs. accumulated dispersion for a0 of 1.2, 1.5 and 1.8 ps, correspond to T0 (full width of half minimum; FWHM) of 2, 2.5 and 3 ps, by Eq. (11). I/O IR shows monotonous decrease by increase of the accumulated dispersion. Therefore, detecting decrease of the I/O IR, waveform distortion by accumulated dispersion is able to be assumed with these relations. These analyses require the information about characteristics of the original signal before distortion, such as pulsewidth and chirp.
2.4 Differential Group Delay
In this section, monitoring technique of differential group delay (DGD) was described. DGD is well known as 1st order PMD, and it is the delay time between two pulses in the two principal state of polarization (PSP) as shown in Fig. 5. DGD and also PSP and polarization splitting ratio (PSR; Ix/Iy in Fig. 4) change frequently. To detect the waveform distortion by DGD such a temporary condition, a periodical rotated polarizer is added to the monitor. The polarizer acts as an analyzer, which couples the pulses in two PSP or x and y axis in Fig. 5. Pulsewidth of analyzed signal is the shortest when the azimuth angle θ correspond with each two PSP. In case of θ is the rest, the pulses in the two PSPs are coupled with slight time difference and the analyzed pulses have wider pulsewidth.
The relation between I/O IR and DGD was calculated using 160 GHz short pulse in the following. Fig. 6(a) illustrates calculated intensity variation with θ in the case of DGD equal to 0, 0.5, 1.0, 1.5 and 2 ps. 2.5 ps-wide chirpless Gaussian pulses were used for the signal pulses and PSR was 1 in this calculation. When pulse widening is only caused by CD, the intensity ratio is constant regardless of θ. On the other hand, when DGD influences the signal distortion, the I/O IR changes by θ. Maximum values are obtained in θ of 0 and π/2 and minimum values are obtained in θ of π/4 and 3π/4. It is obviously shown in Fig. 6(a) that the differences of maximum and minimum values were dependent on DGDs, and the variation indicated DGDs of transmission lines, in which signals are transmitted.
Because PSR is not 1 in many practical cases, the intensity ratio was calculated for various PSR and the results are shown in Fig. 6(b). In this case, θ for minimum values are change, but the intensity ratio variation by θ is same, regardless of the power splitting ratio. Therefore, the intensity ratio variations correspond to the waveform distortion by the DGD.
Fig. 5. Definition of the power splitting ratio Ix/Iy of signal pulse with DGD and azimuthal angle θ of prepositional polarizer.
Fig. 6. Intensity variation with rotation of prepositional polarizer, (a) for each DGDs (a/b=1) and (b) for each PSRs (DGD=1ps).
Figure 7 illustrates that calculated difference of maximum and minimum values with each pulsewidth as a function of DGD and suggests the difference grows with the increase of DGD. Thus, waveform distortion by DGD is able to be evaluated from the detection of intensity variation in case that original pulsewidth are already known. Consequently, I/O intensity ratio of EAM prescaler allows monitoring OSNR, CD and DGD.
2.5 Discrimination of waveform distortion factors
Waveform distortions by DGD are discriminated from the other factors by the intensity variation by the rotating polarizer as explained the previous section. However, it is hard to distinguish accumulated dispersion from OSNR degradation by the I/O IR decrease only. In this section, authors show a technique to discriminate the origins of waveform distortions, such as accumulated CD and OSNR degradation. A Simple block diagram of the monitor to distinguish the distortion origins is illustrated in Fig. 8(a). In the contrast to the basic set-up, which was described in section 5.1, an amplified spontaneous emission (ASE) source optionally added to the input stream to deteriorate the signal waveform in the monitor. These Signals with and without extra ASE were shifted by an optical switch (SW).
Fig. 8. Discrimination technique of waveform distortion factors, (a) schematic diagram of the monitor and I/O IR difference of each case with OSNR degradation (b) and accumulated dispersion (c).
When the extra ASE is added to the signal, the I/O IR curves against the OSNR move parallel along the OSNR axis, and the I/O IR decreases by OSNR degradation are different in each case, as shown in Fig. 8(b). On the other hand, the I/O IR decrease caused by the dispersion is independent from the OSNR. Therefore, the total intensity decrease without (IDta) and with (IDtb) the extra ASE are expressed as
where IDOSNRa and IDOSNRb are the I/O IR decreases by the OSNR degradation with and without the extra ASE, respectively. And IDCD is the decrease by accumulated CD.
Because IDCD is same in each basic monitor as shown in Fig. 8(c), differences of the total intensity decrease (IDta-IDtb) are independent of the waveform distortion by the accumulated CD. Therefore, the I/O IR decreases by the OSNR degradation become measurable from the differences. As a result, the I/O IR decreases by the dispersion are self-evidently determined from (IDta-IDtb) and either decreases by the OSNR degradation (IDOSNRa or IDOSNRb).
3. Experiments
3.1 Verification of the monitoring technique principle
The verification of concepts expressed in previous section was examined using practical 160 Gb/s signals by experimental setup shown in Fig. 9. A 40 GHz pulse train which had 2.5-ps pulsewidth was generated with a mode-locked laser diode (MLLD) [27] and put into an OTDM multiplexer [3]. In the multiplexer, every channel is encoded at 40 Gb/s pseudo random pattern sequence (PRBS) of 215-1 pattern length with four high speed EAMs individually. The carrier phases of each 40 Gb/s OTDM channel modulated signals were optionally adjusted controlling temperature of EAMs by auto temperature controllers (TEC) [28] and output signal is kept carrier suppressed state like Carrier Suppressed Return-to-Zero (CS-RZ) signal [23].
Fig. 9. Experimental setup, (a) is an entire composition; (b) is detail of the DGD emulator and the analyzer, shown as #3 in (a).
160 Gb/s signals input into the demultiplexer with clock extract function [3] and bit error ratio (BER) detector in sequence. A portion of the 160 Gb/s signal was split and input into the prescaled monitor. In the prescaled monitor, the input signal was modulated with 39.75 GHz sinusoidally driven an EA modulator. The EA modulator was designed so that low polarization dependent loss (PDL) of the module was small enough. In this application, PDL of the module was 0.1 dB. For practical reason, we use 0.25 GHz as the offset frequency Δf in the following experiments. Thus a 1 GHz beat signal was output from EA modulator and the beat signal amplified and detected its intensity after the 1 GHz band pass filter. To detect averaged total power of input signal, a portion of the signal is sent to a power meter.
Some short pieces of standard single mode fiber (SSMF) (#1) were inserted between the transmitter and the monitor to generate accumulated CD. An EDFA and a variable attenuator (ATT) (#2) were used to examine the feasibility of the waveform distortion monitor against OSNR degradation. The DGD emulator was composed of polarization controller (PC) using a half wave plate (HWP), a quarter wave plate (QWP) and some short pieces of polarization maintained fibers (PMF) (#3). An analyzer, shown in #3 or Fig. 9(b), was consisted of a HWP, a QWP and a polarizer. It was used instead of the rotating polarizer, which was illustrated in Fig. 5. Another set of an EDFA and an ATT was put before the monitor to add extra ASE for verification of the distinction technique of distortion origins, which was explained in section 2.5. These elements, which were numbered from #1 to #4 were optionally inserted, optimizing following experiments.
First of all, the conversion efficiency of the prescalar was evaluated by the transmitter and shown in Fig. 9. Fig. 10 represents a typical relation between generated 1 GHz and input 160 Gb/s signal. The intensity of x axis was measured averaged input power and y axis shows averaged optical power of a 1 GHz signal by direct modulation. Same electrical intensities were detected in the case that the two optical powers had the relation represented in Fig. 10. As shown in Fig. 10, linear relation is confirmed below +6 dBm. This upper limit of linearity is caused by absorption limit of EAM. Conversion efficiency is 21 dB and sensitivity of a detector used in our experiments is −45 dBm. Therefore, from +6 dBm to −24 dBm of input 160 Gb/s signal is measurable by the beat signal and it corresponds to 30 dB of dynamic range. The results predicted by section 2 suggest that 5 dB of dynamic range is enough to monitor signal distortion. Therefore, minimum tolerant input power to the monitor is about −19 dB in this report.
In the next place, the I/O IR variations by OSNR degradation, accumulated CD and DGD were verified by experimental setup shown in Fig. 9. Fig. 11 shows the I/O IR as the function of the signal OSNR measured by OSA with resolution bandwidth (RBW) of 1 nm. Signal OSNR was initially set 34dB and adjusted using a set of an EDFA and an ATT (#2). Bandwidth of OBPF was 4 nm. Input intensity of the transmitted signal was from 0 dBm to 6dBm. In Fig. 11, theoretical result, which was calculated by Eq. (10), is illustrated by a broken line. The theoretical line clearly corresponds to measured data. Additionally the I/O IRs were equal regardless of input signal intensity. Therefore, I/O IR allows detecting waveform distortion by OSNR degradation. Two eye diagrams with different OSNR are also shown in Fig. 11. They clearly show that the lower OSNR causes more significant distortion of waveform.
By inserting some short pieces of SSMF (#1), which has dispersion of 0.5 ps/nm, impact of accumulated CD to I/O IRs was examined. Fig. 12 shows decrease of I/O IR with accumulated CD. Waveforms are also shown in Fig. 12 for 0 and 3 ps/nm dispersion cases. Calculated results by Eq. (11) and experiment results show a good agreement up to 2 ps/nm of accumulated dispersion. Owing to bit-to-bit interferences of OTDM signals, experimental results are different from the calculated results and decrease of inclination is observed over 2 ps/nm of accumulated CD. Regardless of such a small operating range of CD, it is effective to observe the signal damage by accumulated dispersion, since quality of 160 Gb/s signal is rapidly deteriorated by such a small accumulated CD. In this experiment, 160 Gb/s signals with OSNR of 15~30 dB were used and it was clearly shown that OSNR did not influence I/O intensity ratio. These results support that the technique with switching extra optical noise, which was shown in Fig. 8, enables to detect signal distortions by OSNR degradation and accumulated dispersion separately.
Fig. 11. I/O IR vs. OSNR degradation. Broken line is calculated result by Eq. (8). Input power is from 0 to 6 dBm. Eye-diagrams with OSNR of 30 dB and 15 dB are illustrated in the right side as examles.
Fig. 12. I/O IR decrease according to accumulated dispersion. Broken line is calculated result by Eq. (9). OSNR of the input signal is from 15 to 30 dB. Eye-diagrams with CD of 0 and 3 ps/nm are illustrated in the right side as examles.
To verify the effectiveness of the monitoring technique against DGD, the relation of I/O IR and DGD were examined by set-up shown in Fig. 9(b). DGD generated by the emulator (#3) were measured by an optical sampling oscilloscope with a polarization diversity sampling head [29]. Example of measured waveform is shown in Fig. 13(a). Ix and Iy are signals of parallel to x- and y-axis in Fig. 5. In Fig. 13(a), DGD was observed by offset time of pulse peaks between Ix and Iy. Itotal is mixed signal of Ix and Iy and observed damaged signal by DGD. The measured waveforms were parts of PRBS signal with the bit-length of 27-1. It was easy to measure DGD from delay of two signals in each PSP (Ix and Iy). In this case, DGD was estimated as 1.1 ps from the waveforms.
By emulating DGD and state of polarization (SOP) using the equipment, impact of I/O IR by DGD was verified. Fig. 13(b) shows difference of maximum and minimum values of I/O IR vs. various DGDs by replacing pieces of PMF. Variations of I/O IR increase according to signal distorted by larger DGD and this result reflect the prediction shown in Fig. 7. In this figure, experimental and simulated results show good agreement.
Fig. 13. DGD measurement by the novel method, (a) example of DGD measurement using an optical sampling scope [29] and (b) differences of maximum and minimum values of I/O IR.
Verification of the distortion origins discrimination, which was described in Fig. 8, was performed with the set-up shown in Fig. 9. To deteriorate the signal waveform with extra optical noises in the monitor, one set of an EDFA and an ATT (#4) was inserted. OSNR of input signals were set 15 dB and 17 dB by adding attenuation of 15 dB and 16 dB before the EDFA. As same as above mentioned, RBW of OSA was set 1 nm. I/O IRs of the signals with extra noises were measured and compared with the object signal, of which initial OSNR was set at 34 dB. The differences of I/O IRs were shown in Fig. 14(a). Triangles and diamonds are the difference in case OSNR of 15 dB and 17 dB, respectively. Considering I/O IRs were not change by OSNR under the condition that accumulated CD is same as shown in Fig. 8, these differences indicate the OSNR degradation. In both cases with additional optical noise, the I/O IR rapidly decreased below a particular OSNR, and the operating ranges were 5 dB in these experiments by the limitation of the sensitivity of the detector. Therefore, the technique is suitable of OSNR measurement with high resolution but narrow operating range.
Fig. 14. I/O IR different of dual configuration monitoring technique, (a) results (b) experimental set-up using dual filters.
Another technique to vary the signal OSNR was tested in these experiments. OBPFs with broad bandwidth, such as 10 nm and 15 nm, were used instead of a set of the EDFA and the variable ATT. The set-up was illustrated in Fig. 14 (b). The results were illustrated in Fig. 14(a), for the comparison with the technique previously mentioned. Squares and circles show the result in cases of filter bandwidth Δλ (Fig.14 (b)) of 10 nm and 15 nm, respectively. In case of the technique using the filter with wider bandwidth, the difference increases gradually by the decrease of OSNR and the technique is suitable for measurement with wide operating range, though with low resolution.
Figure 15 illustrates measured I/O IRs and the difference of signals after field transmission of 127 km installed fiber in Japan Gigabit Network (JGN)-II testbed [24] over 120 minutes. The I/O IRs were detected every 10 second. The difference was given by two monitors, which had filters with different Δλ2, such as 4 nm and 15 nm. The difference was kept within +/− 0.02 dB and this result indicates OSNR of the signal was not degraded during this observation. On the other hand, the I/O IRs of the transmitted signal decreased 1.5 dB because signal pulses were widen by the dispersion. 1.5 dB decrease of I/O IR correspond to 2.7 ps-wide pulses were broaden to 3 ps-wide by chromatic dispersion of 1.2 ps/nm or DGD of 1 ps. In this experiment, it was considered that 160 Gb/s was mainly damaged by DGD because the degree of polarization (DOP) decreased to 90 % correspond to DGD of 1 ps during this observation [24].
Fig. 15. Measured I/O IR and the difference of signals after transmission of 127 km installed fiber in JGN-II testbed [24] over 120 minutes.
3.2 Q factor estimation by the monitoring technique
Finally we examined relation between signal qualities and I/O IR decreases by experiments using the equipment shown in Fig. 9. The signal quality was evaluated by Q factors, which were estimated from the decision voltage vs. BER characteristics [30]. At back-to-back measurement point, Q factors of about 27 dB were obtained on each demultiplexed 40 Gb/s signal. Fig. 16 (a) represents the relation between I/O IR decrease and Q penalty according to OSNR degradation. The signal OSNR was adjusted by the set of the EDFA and the attenuator (#2 in Fig. 9 (a)). Initial OSNR of signals with or without extra noise by #4 in Fig. 8 (a) was 34 dB and 17 dB, respectively.
As shown in Fig. 16 (a), Q penalty have close relation to I/O IR, and Q penalty is able to be estimated from I/O IR under the condition that the relations are already known. Furthermore, Fig. 16 (a) shows that the resolution of the monitor is lower in case with higher OSNR signal, and the resolution is higher in case with lower OSNR signal contrasting it. Therefore, the resolution of the proposed monitor is able to be improved by deteriorated with extra optical noise even in case with measuring higher OSNR signals. In this experiment, the resolution of the monitor was 0.1 dB reflecting the residual PDL of the prescalar using EA modulator. Therefore, assuming the Q factor of signal under test is 27 dB, 2 dB of Q penalty is able to be identified with extra optical noise. Considering the OSNR degradation is actually caused due to the breakdown of EDFAs, the OSNR degradations of the transmitted signals are very large, and it is effective to estimate roughly signal quality degradation by I/O IR decrease. Fig. 16 (b) shows the relation between I/O IR decrease and Q penalty according to accumulated CD. Because the proposed monitor has high resolution for accumulated CD, the penalty of signal quality by increase of accumulated CD was large enough against the monitor resolution.
Fig. 16. Relation between I/O IR and Q penalty for signals with OSNR degradation (a) and for signals with distortion by accumulated dispersion (b).
4. Conclusion
We explained detailed principle of the monitoring technique of waveform distortion for ultra high speed signal, such as 160 Gb/s, by the clock tone detection with the asynchronous prescalar using an EA modulator. The proposed technique has applicability to monitor signal distortion by OSNR degradation and CD and DGD. Assuming limit of electrical circuit is 50 GHz, possibility of monitoring up to 400 Gb/s signal is shown by simulated results. The frequency of prescaled signal should be more than 250 MHz considering +/-20 MHz of oscillator stability, 50 dB of sideband suppression, and 0.01 dB of flatness through the passband of the filter using after o/e conversion. In experiments, a 160 Gb/s signal was downconverted to 1GHz by 39.75 GHz sinusoidally driven EA modulator. The conversion efficiency of the beat signal was about 21 dB and the dynamic range of the monitor is from +6 dBm to −24 dBm. By inserting prepositional noise source and switch, the waveform distortions by OSNR degradation and CD are detected independently with the monitor. DGD are detected by the intensity variation in accordance with rotation of signal polarization. From experiments and numerical estimations, we also verified that the efficiency of performance of the proposed monitoring technique. The operating range of CD was narrow in case of OTDM 160 Gb/s signal because of the pulse-to-pulse carrier phase interaction. But it was enough, since signal quality is rapidly deteriorated by accumulated CD. In addition, observation of 127 km transmitted signal distortion by the monitor was demonstrated over 120 minutes using an installed fiber in JGN-II testbed. The I/O IR was decrease 1.5 dB and the dominant factor of the signal distortion is considered DGD from the experiments. Finally we demonstrated signal quality monitoring using the technique by experiments. The signal quality is rapidly deteriorated by accumulated CD and it was possible to detect with high accuracy. On the other hand, because of lower sensitivity against OSNR degradation and the slight PDL of EA modulator in these experiments, 2 dB penalty form 27 dB of Q factor was identified.
Acknowledgments
A part of this work is supported by National Institute of Information and Communications Technology (NICT) of Japan. And we thank Prof. K. Kitayama and Prof. A. Maruta from Osaka University for significant discussions about monitoring techniques.
References and links
1. M. Nakazawa, T. Yamamoto, and K. R. Tamura, “1.28 Tbit/s-70 km OTDM transmission using third- and fourth-order simultaneous dispersion compensation with a phase modulator,” Electron. Lett. 36, 2027–2029 (2000). [CrossRef]
2. H. G. Weber, S. Ferber, M. Kroh, C. Schmidt-Langhorst, R. Ludwig, V. Marembert, C. Boerner, F. Futami, S. Watanabe, and C. Schubert, “Single channel 1.28 Tbit/s and 2.56 Tbit/s DQPSK transmission,” Electron. Lett. 43, 178–179 (2006). [CrossRef]
3. H. Murai, M. Kagawa, H. Tsuji, and K. Fujii, “Single channel 160 Gbit/s carrier-suppressed RZ transmission over 640 km with EA modulator based OTDM module,” Proc. 30th European Conf. on Opt. Commun. (ECOC2003) Mo3.6.4 (2003).
4. E. Lach, K. Schuh, M. Schmidt, B. Junginger, G. Charlet, P. Pecci, and G. Veith, “7×170 Gbit/s (160 Gbit/s + FEC overhead) DWDM transmission with 0.53 bit/s/Hz spectral efficiency over long haul distance of standard SMF,” in Proc. 29th European Conf. on Opt. Commun. (ECOC2003) Th4.3.5 (2003).
5. S. Kawanishi, H. Takara, K. Uchiyama, I. Shake, and K. Mori, “3 Tbit/s (160 Gbit/s ×19ch) OTDM-WDM transmission experiment,” in Tech. Dig.Optical Fiber Communications Conf. 1999 (OFC’99) PD1 (1999). [CrossRef]
6. A. Suzuki, X. Wang, T. Hasegawa, Y. Ogawa, S Arahira, K. Tajima, and S. Nakamura, “8×160 Gb/s (1.28 Tb/s) DWDM/OTDM unrepeatered transmission over 140 km standard fiber by semiconductor-based devices,” in Prc. 29th European Conf. on Opt. Commun. (ECOC2003) Mo3.6.1 (2003)
7. Y. Takushima and K. Kikuchi, “In-service monitor for group-velocity dispersion of optical fibre transmission systems,” Electron. Lett. 37, 743–745 (2001). [CrossRef]
8. G. Bendelli, C. Cavazzoni, R. Girardi, and R. Lano, “Optical performance monitoring techniques,” in Proc. 26th Euro. Conf. Opt. Commun. (ECOC2000) 4, 113–116 (2000).
9. S. Norimatsu and M. Maruoka, “Accurate Q-factor estimation of optically amplified systems in the presence of waveform distortion,” IEEE J. Lightwave Technol. 20, 19–27 (2002). [CrossRef]
10. T. E. Dimmick, G. Rossi, and D. J. Blumenthal, “Optical monitoring technique using double sideband subcarriers,” Photon. Technol. Lett. 12, 900–902 (2000). [CrossRef]
11. B. Kozicki, A. Maruta, and K. Kitayama, “Experimental demonstration of optical performance monitoring for RZ-DPSK signals using delay-tap sampling method,” Opt. Express 16, 3566–3576 (2008). [CrossRef] [PubMed]
12. I. Shake, H. Takara, and S. Kawanishi, “Simple Q factor monitoring for BER estimation using opened eye diagrams captured by high-speed asynchronous electrooptical sampling,” Photon. Technol. Lett. 15, 620–622 (2003). [CrossRef]
13. B. Kozicki, T. Ohara, and H. Takara, “Optical Performance monitoring of phase-modulated signals using asynchronous amplitude histogram analysis,” IEEE J. Lightwave Technol. 26, 1353–11361 (2008). [CrossRef]
14. S. D. Dods and T.B. Andersion, “Optical performance monitoring technique using delay tap asynchronous waveform sampling,” in Tech. Dig. Optical Fiber Communications Conf. 2006 (OFC’06) OThP5 (2006).
15. T. B. Anderson, S. D. Dods, E. Wong, and P. M. Farell, “Asynchronous measurement of chromatic dispersion from waveform distortion,” in Tech. Dig. Optical Fiber Communications Conf. 2006 (OFC’06), OWN4 (2006).
16. K. T. Tsai and W. I. Way, “Chromatic dispersion monitoring using an optical delay-and-add filter,” IEEE J. Lightwave Technol. 23, 3737–3747 (2005). [CrossRef]
17. A. L. Campillo, “Chromatic dispersion-monitoring technique based on phase-sensitive detection,” Photon. Technol. Lett. 17, 1241–1243 (2005). [CrossRef]
18. Z. Pan, Q. Yu, Y. Xie, S. A. Havstad, A. E. Willner, D. S. Starodubov, and J. Feinberg, “Chromatic dispersion monitoring and automated compensation for NRZ and RZ data using clock regeneration and fading without adding signaling,” in Tech. Dig. Optical Fiber Communications Conf. 2001 (OFC’01) WH5-1 (2001).
19. X. Liu and Y. H. Kao, “A simple OSNR monitoring technique independent of PMD and chromatic dispersion based on a 1-bit delay interferometer,” Proc. 32th European Conf. on Opt. Commun. (ECOC2006) Mo4.4.7 (2006). [CrossRef]
20. Y. K. Lize, L. Christen, J. Y. Yang, P. Saghari, S. Nuccio, A. E. Willner, and R. Kashyap, “Independent and simultaneous monitoring of chromatic dispersion and polarization-mode dispersion in OOK and DPSK transmission,” Photon. Technol. Lett. 19, 3–5 (2007). [CrossRef]
21. S. Wielandy, M. Fishteyn, T. Her, D Kudelko, and C. Zhang, “Real-time measurement of accumulated chroamtic dipersion for automatic dispersion compensation,” Electron. Lett. 38, 1198–1199 (2002) [CrossRef]
22. M. Kagawa, H. Murai, H. Tsuji, and K. Fujii, “Waveform Distortion Monitor for 160 Gbit/s Signal by Prescaled-clock Measurement using EA Modulator,” in Tech. Dig.Optical Fiber Communications Conf. 2008 (OFC’08) OThW7 (2008).
23. Y. Miyamoto, A. Hirano, K. Yonenaga, A. Sano, H. Toba, K. Murata, and O. Mitomi, “320 Gbit/s (8×40 Gbit/s) WDM transmission over 367-km zero-dispersion-flattened line with 120-km repeater spacing using carrier-suppressed return-to-zero pulse format,” Optical Amplifiers and Their Applications 1999, PDP4 (1999).
24. Y. Kanda, H. Murai, M. Kagawa, and K. Fujii “Highly stable 160-Gb/s field transmission employing adaptive PMD compensator with ultra high time-resolution variable DGD generator,” Proc. 34th European Conf. on Opt. Commun. (ECOC2008) We3.E.6 (2008). [CrossRef]
25. M. E. Valkenburg, Analog Filters Design. Philadelphia: Holt and Rinehart and Winston (1982).
26. G. P. Agrawal, Nonlinear Fiber Optics. New York: Academic (1989).
27. S. Arahira and Y. Ogawa, “40 GHz actively mode-locked distributed Bragg reflector laser diode module with an impedance-matching circuit for efficient RF signal injection,” Jpn. J. Appl. Phys. 43, 1960–1964 (2004). [CrossRef]
28. M. Kagawa, H. Murai, H. Tsuji, K. Sasaki, and K. Fujii, “Control and Stabilization of bit-wise phase correlation in 160 (4×40) Gbit/s OTDM signal and its impact on transmission,” Opt. Express 16, 10039–10052 (2008). [CrossRef] [PubMed]
29. A. Otani, Y. Tsuda, and K. Igawa, “Development of Polarization Diversity Sampling Head Unit using Organic AANP Crystals and 160Gb/s Far-end Waveform Measurement,” IEEJ Trans. FM 126, 409–413 (2006). [CrossRef]
30. S. Bergano, F. Kerfoot, and C. R. Davidson, “Margin measurements in optical amplifier systems,” Photon. Techenol. Lett. 5, 304–306 (1993). [CrossRef]
