Optical performance monitoring techniques based ... - OSA Publishing

Optical performance monitoring techniques based on pilot tones for WDM network applications H. C. Ji, K. J. Park, J. H. Lee, H. S. Chung, E. S. Son, K. H. Han, S. B. Jun, and Y. C. Chung Korea Advanced Institute of Science and Technology, Department of Electrical Engineering 373-1, Guseong-dong, Yuseong-gu, Daejeon 305-701, Korea [email protected] R ECEIVED 1 D ECEMBER 2003; REVISED 18 M ARCH 2004; ACCEPTED 30 M ARCH 2004; PUBLISHED 14 J UNE 2004

We report on the pilot-tone-based optical performance monitoring techniques for use in the wavelength-division multiplexing (WDM) network. We first described the operating principle of the pilot-tone-based monitoring technique and estimated its scalability by analyzing potential problems. We then reviewed the techniques proposed for monitoring various optical parameters of WDM signals—such as optical power, wavelength, optical path, and cross talk—using pilot tones. We also presented examples of various network elements using the pilot-tone-based monitoring technique. In addition, we reviewed the pilottone-based monitoring techniques used in adaptive compensators for chromatic dispersion and polarization-mode dispersion, and we discussed their limitations and possible solutions. For example, we demonstrated a simple technique for monitoring the chromatic dispersion and polarization-mode dispersion simultaneously, without their affecting each other, by scrambling the state of polarization of the optical signal. © 2004 Optical Society of America OCIS codes: 060.4510, 060.4250.

1.

Introduction

Recently, the capacity of the fiber-optic network has been increased drastically with wavelength-division multiplexing (WDM) technology. For the proper operation and maintenance of such a high-capacity network, it is essential to monitor the signal’s quality. Previously, this was mostly achieved by use of the synchronous optical network (SONET)/synchronous digital hierarchy (SDH) layer. However, as the amplified transmission distance becomes longer, there is less opportunity to monitor the signal’s quality by use of the protocol layer. In addition, it is envisioned that in the near future, WDM networks would be able to add–drop or cross-connect optical signals without the optical-to-electronic (O/E) conversions. Under these circumstances, it would be inevitable to monitor the signal’s quality directly in the optical layer [1]. There have been many efforts to use pilot tones (i.e., small sinusoidal components added to WDM signals) for the monitoring of WDM signals directly in the optical layer. For example, it has been reported that pilot tones could be used to monitor various optical parameters of WDM signals such as optical power, wavelength, and optical signal-to-noise ratio (OSNR), and so on [2–7]. The pilot-tone-based techniques could be used to monitor these parameters without the expensive demultiplexing filters (such as tunable optical filter and diffraction grating). Thus this technique could be extremely cost-effective. In addition, this technique is well suited for use in a dynamic WDM network, since the pilot tones are bound to follow their corresponding optical signals anywhere in the network. Thus the

© 2004 Optical Society of America JON 3468 July 2004 / Vol. 3, No. 7 / JOURNAL OF OPTICAL NETWORKING

510

optical path of each WDM signal could be monitored simply by tracking its tone frequency [8–11]. Although the pilot-tone-based monitoring technique has many advantages, it also has some limitations owing to the following problems. First, the pilot tone could impose unwanted amplitude modulation on the data signal and degrade the receiver sensitivity [12]. Second, the performance of the pilot-tone-based monitoring technique could be deteriorated by ghost tones caused by cross-gain modulation (XGM) and stimulated Raman scattering (SRS) [13]. These problems could be mitigated by using proper amplitudes and frequencies of pilot tones. However, for use in a long-haul network with a large number of channels, it may still be necessary to restrict the number of WDM channels to be monitored at a time (by using an optical bandpass filter). Pilot tones could also be used to monitor the chromatic dispersion (CD) and polarization-mode dispersion (PMD) for adaptive compensators [7, 14–17]. These techniques typically used high-frequency (>1 GHz) pilot tones. However, the main obstacle for these techniques is the difficulty in separating their effects, since the magnitude of such high-frequency pilot tones is dependent on both types of dispersions [18]. To solve this problem, we proposed use of the phase-modulated (PM) pilot tones, instead of the conventional amplitude-modulated (AM) pilot tones, for monitoring CD [18]. For the PMD monitoring technique, we could use the single sideband (SSB) pilot tones to suppress the effect of CD [19]. We also showed that both CD and PMD could be monitored simultaneously, without affecting each other, by polarization scrambling the optical signal [20]. In this paper, we review the pilot-tone-based monitoring technique and its potential problems. In Section 2, we describe the operating principle of the pilot-tone-based monitoring technique and estimate the scalability of this technique by analyzing its limitations. Using this analysis, we also determine the proper frequencies and amplitudes of pilot tones. We then review the techniques for monitoring various optical parameters of WDM signals such as optical power, wavelength, optical path, and cross talk with pilot tones. In addition, we present examples of various network elements using the pilot-tone-based monitoring technique. In Section 3, we review the pilot-tone-based monitoring techniques for CD and PMD used in adaptive compensators. We first review the monitoring techniques for CD and compare their performances by considering the effects of self-phase modulation (SPM) and PMD. We also review the PMD monitoring technique on the basis of pilot tones and discuss its performance in the presence of CD. We then introduce several novel techniques for monitoring CD and PMD without their affecting each other. Finally, this paper is summarized in Section 4. 2.

Optical Performance Monitoring Techniques Based on Pilot Tones

In this section, we describe the operating principle of the pilot-tone-based optical performance monitoring technique and discuss its problems and possible solutions. We then review the techniques for monitoring various optical parameters of WDM signals such as optical power, wavelength, optical path, and cross talk with use of pilot tones. In addition, we estimate the scalability of these techniques and use the information to determine the proper range of tone frequencies for the targeted network. 2.A.

Operating Principle

Figure 1 shows the operating principle of the pilot-tone-based monitoring technique [8]. We assume that an optical signal is transmitted from node A to node C via node B. A small sinusoidal component (i.e., pilot tone) is added to the optical signal at node A. This pilot tone can be extracted at node B by use of a simple electronic circuit and can be used for monitoring various optical parameters such as optical power, wavelength, OSNR, and so


511

on. Pilot tones can also be used to monitor the optical paths of WDM signals even in a dynamically reconfigurable network. This is because once the pilot tone is attached, it is bound to follow the optical signal throughout the network. Thus we can monitor the optical path of each WDM signal simply by tracking its corresponding tone frequency. pilot tone

pilot tone

node A

node B

signal

node C

Fig. 1. Pilot-tone-based optical Fig. 1performance monitoring technique [8].

For practical applications, pilot tones should be added into and extracted from the optical signal anywhere in the network. Figure 2 shows typical techniques used for generating and detecting pilot tones. A pilot tone could be generated by dithering the laser’s bias current [see Fig. 2(a)], the bias voltage of amplitude modulator [see Fig. 2(b)], or the phase modulator [see Fig. 2(c)]. These techniques would require a slight modification of the existing transmitter. However, it should be noted that the pilot tones could also help suppress stimulated Brillouin scattering [21]. For the detection of pilot tones, we proposed use of the technique which is based on the fast Fourier transform (FFT), which is shown in Fig. 2(d). This technique is attractive, since all the amplitudes and frequencies of pilot tones added to WDM signals could be measured simultaneously without any scanning mechanism. However, it would be difficult to use this technique when the tone frequency is higher than several hundreds of megahertz. In such cases, a tunable electrical bandpass filter or a tunable local oscillator could be used, as shown in Figs. 2(e) and 2(f), respectively.

Bias

LD

LD Bias

AM

LD

AM Data

Bias

PM Bias

Data

(a)

(b)

(c)

PD

PD

PD

A/D

BPF

FFT

RFD

(d)

(e)

LOSC RFD

(f)

Fig. 2 methods: (a) adding a small sinusoidal current Fig. 2. Pilot tone generation and detection to the laser’s bias current, (b) dithering of the bias voltage of external modulator, (c) PM tone generation by use of a phase modulator, (d) pilot-tone detection using FFT, (e) using tunable electrical bandpass filter, (f) using a tunable local oscillator for the down-conversion of tone frequency (LD, laser diode; AM, amplitude modulator, PM, phase modulator; PD, photodetector; A/D, analog-to-digital converter; FFT, fast Fourier transform; BPF, tunable bandpass filter; RFD, rf power detector; LOSC, tunable local oscillator).


512

2.B.

Potential Problems

When the pilot tone is added to the optical signal, it could interfere with the spectral components of data and cause power penalty. Thus it is necessary to limit the maximum modulation index (MI) of pilot tone to maintain the power penalty within the acceptable level [12]. However, owing to the frequency response characteristics of the receiver, the maximum allowable MI is also dependent on the tone frequency. For example, Fig. 3 shows the power penalties of 10-Gbit/s NRZ signal (pattern length, 231 − 1) measured while varying the MI and frequency of the pilot tone. The result shows that the power penalty could be substantially reduced by use of a tone frequency lower than ∼100 kHz. This was because the receiver used in this experiment had a low-frequency cut-off at ∼150 kHz. Thus, the power penalty could be maintained within 0.5 dB even when the MI was as high as 12%. However, the MI should have been set to be smaller than 6% when we used the tone frequency in the range of approximately 1 MHz to 4.5 GHz. When the tone frequency was higher than 4.5 GHz, the pilot-tone induced power penalty decreased because of the roll-off characteristics of our 10-Gbit/s receiver.

Power Penalty (dB)

14

MI = 12 %

1.0

2.0

0.8 0.6

1.5

0.4 0.2 0.0

1.0

B.W. = 150 kHz 1

10

100

1000

MI = 6 % 0.5 0.0

MI = 3 % 1

10

100

1000

Tone Frequency (kHz)

(a)

10000

RF tone

power

Receiver Response

1.2

MI of RF Tone (%)

2.5

12 10

Data spectrum frequency

8 6 4

Due to LPF characteristics of receiver

2 0

1

2

3

4

5

6 7 8 9 10

Tone Frequency (GHz)

(b)

Fig. 3. Effects of modulation index (MI) Fig. 3and frequency of pilot tone on 10-Gbit/s NRZ signal (pattern length, 231 − 1): (a) pilot-tone induced power penalty measured at low tone frequencies, (b) the maximum allowable modulation indices of high-frequency pilot tones for 0.5-dB penalty.

When the pilot-tone-based monitoring technique is used in an amplified WDM network, its performance could be deteriorated by the cross-gain modulation (XGM) of an erbium-doped fiber amplifier (EDFA) and/or stimulated Raman scattering (SRS) [13, 22]. Figure 4 illustrates the mechanisms of these performance degradations. When WDM signals are slightly modulated with low-frequency sinusoidal components (i.e., pilot tones), the EDFA’s gain becomes also modulated at the same frequencies owing to its slow gain dynamic property. This XGM could generate unwanted cross-talk components at the WDM channels [22]. We designated this cross-talk component as ghost tone. These ghost tones could not only cause measurement errors of the monitoring technique, but also mislead the network operators (to interpret the dropped channels as still existing). This problem could be solved simply by using high-frequency (>1 MHz) pilot tones [22] and/or automatic gain-controlled EDFAs [13]. In this case, however, the performance could be impaired by the interactions between the optical signals and pilot tones via SRS [13]. To exemplify these problems, we performed a long-distance WDM transmission experiment using eight distributed-feedback (DFB) lasers operating at approximately 1547.72 –1558.98 nm (channel spacing: 200 GHz). Each laser was slightly dithered by the modulating of its bias current


513

with a small sinusoidal component ranging from approximately 101–115 kHz (separation, 2 kHz; MI, 10%). The transmission link consisted of 640 km of single-mode fiber (SMF) and eight EDFAs. To evaluate the effect of XGM, we dropped six out of eight WDM channels after transmission over 640 km of SMF and measured the optical spectrum of WDM signals and electrical spectrum of their corresponding pilot tones. fiber

EDFA

XGM

optical carrier 1 f1

optical carrier 2

pilot tone f1 f2

SRS

optical carrier 1

f2 freq

f2 f1 f1 f2

optical carrier 2

optical carrier 1

f2 f1 f1 f2

ghost tones

freq

f2 f1

f1

f2

SRS

optical carrier 2 f2

f1 f1 f2

ghost tones

freq

Fig. 4. The mechanisms of performance Fig. 4 degradation caused by XGM and SRS.

Figure 5 shows the results. Although the optical spectrum in Fig. 5(a) shows that six channels were completely dropped, the electrical spectrum of pilot tones in Fig. 5(b) shows the ghost tones generated by XGM at the frequencies corresponding to the dropped channels. To suppress these ghost tones, we applied a dynamic gain control unit (using an additional laser) in the first EDFA of the transmission link [23]. In this case, however, we turned off the fifth WDM channel operating at the same wavelength (1554.13 nm) with the control channel. This was to maintain identical total input power of the EDFA. Figure 5(c) shows the electrical spectrum of pilot tones measured at the output of the first EDFA. The ghost tones were removed completely as we compensated the effect of XGM using the control channel. However, when we measured the electrical spectrum after transmission over 640 km of SMF, the ghost tones reappeared, as shown in Fig. 5(d). The amplitudes of these ghost tones were measured to be larger for the channels separated further from the control channel and increased proportional to the square of the signal power. Thus, we concluded that these ghost tones were caused by the SRS-induced cross talk. Figure 6(a) shows that with pilot tones, each WDM channel can be represented by a distinct pilot tone in the electrical spectrum. However, as WDM signals traverse through an optical network, these pilot tones could generate ghost tones on the other WDM channels via XGM and SRS, as shown in Fig. 6(b). These ghost tones could cause measurement errors if we tap a small portion of light from the transmission fiber and detect the pilot tones of all WDM channels with use of a single photodiode, as shown in Fig. 6(c). However, this problem could be avoided by demultiplexing the WDM channels before the detection of the pilot tones, as shown in Fig. 6(d). This is because the demultiplexed WDM signal does not have any ghost tone (caused by other WDM channels) at its corresponding tone frequency. 2.C.

Scalability

The pilot-tone-based monitoring technique is extremely cost-effective, since we could tap a small portion of light from the transmission fiber and use it for the performance monitoring of every WDM signal without using the demultiplexing filter. However, in this case, the


514

10 dB/div

5 dB/div dropped signals

ghost tones

1562.1 nm

1546.1 nm

118 kHz

98 kHz

(a)

(b)

10 dB/div

10 dB/div ghost tones

no ghost tones

control channel

98 kHz

118 kHz

(c)

98 kHz

118 kHz

(d)

Fig. 5. Measured optical and electrical (a) optical spectrum measured after 640Fig.spectra: 5 km transmission, (b) electrical spectrum measured after 640-km transmission (without using control channel), (c) electrical spectrum measured after the first EDFA (using control channel), (d) electrical spectrum measured after 640-km transmission (using control channel).


515

accuracy of the pilot-tone-based monitoring technique could be seriously affected by the ghost tones caused by XGM and SRS. Thus we investigated the maximum size of optical network that the pilot-tone-based monitoring technique could support. For this estimation of scalability, we calculated the maximum number of EDFA spans as a function of the number of WDM channels. We assumed that the rf power of ghost tone should be at least 10 dB smaller than that of the original pilot tone. This was to avoid any misinterpretation of the dropped and passed channels by the network operators. Under this condition, the ghost-tone-induced measurement error for optical power would be less than 0.5 dB. We also assumed that the channel spacing was 100 GHz and the optical power of each channel was 3 dBm. The results are shown in Fig. 7. The solid and dashed curves represent the limitations imposed by the SRS and XGM, respectively [13]. The effect of XGM appears to be dominant when the tone frequency is low (10 Gbit/s) channels. In such a high-speed network, the most important limiting factors would be CD and PMD. To overcome these limitations, various dispersion compensators have been proposed and demonstrated. However, in a dynamic network, a WDM channel could experience different CD and PMD whenever the network is reconfigured. In addition, both CD and PMD are sensitive to ambient temperature [30, 31]. Thus, for the efficient compensation of CD and PMD, we should be able to monitor them accurately. Previously, it has been reported that CD and PMD could be monitored with high-frequency pilot tones [7, 14–17]. However, the main obstacle in these techniques is the difficulty in differentiating the effects of CD and PMD. In this section, we review the CD and PMD monitoring techniques based on pilot tones and discuss their limitations and possible solutions. 3.A.

Chromatic Dispersion Monitoring

Recently, it has been reported that the CD of each WDM channel could be monitored by measuring either the magnitude of the AM pilot tone [7, 16] or the magnitude of the AM component converted from the PM pilot tone (i.e., PM–AM conversion) [17]. However, these pilot-tone-based monitoring techniques could be affected by both SPM and PMD [18]. Thus, in this section, we compare the performance of these CD monitoring techniques based on the high-frequency pilot tones by considering SPM and PMD. 3.A.1.

Principles

When an optical signal with AM or PM pilot tones is propagated along optical fiber, the phase difference between the lower and upper sidebands could be changed by CD [7, 16, 17]. As a result, the magnitude of the transmitted pilot tone would be changed. Thus, we could monitor the CD of WDM channel by measuring the magnitude of the received AM or PM pilot tone. The electrical powers of the received AM (PAM ) and PM pilot tones (PPM ) could be described as πDLλ2 f 2 PAM ∝ m2 cos2 (AM pilot tone) , (1) c πDLλ2 f 2 PPM ∝ J02 (A) J12 (A) sin2 (PM pilot tone) , (2) c where c is the speed of light, λ is the wavelength, m and f are MI and tone frequency, D and L are the dispersion parameter and fiber length, A is the depth of phase modulation, and J0 and J1 are the first-kind Bessel functions of order 0 and 1, respectively. Equation (1) shows that the magnitude of AM pilot tones decreases as the fiber length increases. However, if the tone frequency f were too low, the magnitude of the AM pilot tone would remain nearly as a constant. The resolution of this technique [i.e., ∆PAM /∆ (DL)] could be improved by increasing the tone frequency, but this would reduce the measurement range. Thus, the tone frequency should be optimized to obtain both reasonable resolution and measurement range. The dispersion of 1000 ps/nm would cause a 1-dB power penalty for 10-Gbit/s NRZ signals. Our calculation shows that when the tone frequency is 8 GHz the measurement range is given by 0–1000 ps/nm. Thus, in most previous reports, a highfrequency tone (∼8 GHz) was used for the monitoring of CD [7, 16]. On the other hand, Eq. (2) shows that the magnitude of the AM component generated by the PM–AM conversion would increase with CD even when the tone operates in the low-frequency region. Thus, unlike the techniques using AM pilot tones, it would be possible to monitor the value of CD by using a relatively low-frequency (∼2 GHz) PM pilot tone.


524

Figure 15(a) shows the experimental setup used to compare the performance of the dispersion monitoring techniques using AM and PM pilot tones. The output power and operating wavelength of the DFB laser were set to be 3 dBm and 1550 nm, respectively. The output of the DFB laser was modulated at 10 Gbit/s (pattern length, 231 − 1) by using a LiNbO3 intensity modulator. We added an AM pilot tone to the 10-Gbit/s NRZ signal with use of an additional LiNbO3 intensity modulator (or a PM pilot tone with use of an additional LiNbO3 phase modulator). The output signal of the transmitter was sent to the SMF via an EDFA and a variable attenuator. After transmission, the electrical power of the received pilot tone was measured with an electrical spectrum analyzer. The signal power incident on the photodetector was adjusted to be −10 dBm by use of another variable attenuator. Figure 15(b) shows the modulation indices of AM and PM pilot tones measured at the receiver while varying CD (i.e., using different length of fiber) in comparison with the theoretically calculated lines. The frequencies of the AM and PM pilot tones were set to be 8 and 2 GHz, respectively. The depth of phase modulation was set to be 0.09π. The results show that both techniques provide a resolution of approximately 0.006% ps/nm. However, unlike the technique using high-frequency (8 GHz) AM pilot tones, the measurement range could be extended to >15, 000 ps/nm by using the relatively low-frequency (2 GHz) PM pilot tone.

LD

IM or PM

IM

Power meter

BPF SMF PD

PRBS 10 Gb/s 231-1

Signal generator RFSA

(a) 12 PM pilot tone (2 GHz) AM pilot tone (8 GHz) calculation

Modulation Index (%)

10 8 6 4 2 0

0

400

800

1200

1600

2000

Chromatic Dispersion (ps/nm) (b)

Fig. 15. (a) Experimental setup to measure Fig. 15 the CD, (b) MI of pilot tones measured at the receiver while varying the CD (RFSA, rf spectrum analyzer).

3.A.2.

Effects of Self-Phase Modulation and Polarization-Mode Dispersion

When the high-frequency pilot tone is used, the performance of the CD monitoring technique could be deteriorated by SPM [18]. This is because the AM component of the pilot tone could modulate the refractive index of optical fiber and cause phase modulation owing


525

800

100

Monitoring Error (ps/nm)

Monitoring Error (ps/nm)

to SPM. The SPM-induced phase modulation would then generate an additional AM component owing to PM–AM conversion in the dispersive fiber, which, in turn, would cause monitoring errors. Figure 16(a) shows the SPM-induced monitoring errors measured while varying the signal power incident on the 60-km-long SMF (dispersion, 1000 ps/nm), in comparison with theoretically calculated curves [32]. When the signal power incident on the SMF was 10 dBm, the monitoring errors of the techniques using AM and PM pilot tones were measured to be −266 and 18 ps/nm, respectively. Thus, the dispersion monitoring technique using AM pilot tones was much more sensitive to SPM than the technique using PM pilot tone. This was mainly because we set the frequency of AM pilot tones (8 GHz) to be much higher than that of PM pilot tones (2 GHz) to obtain similar resolutions. In general, the effect of SPM increases with the square of the tone frequency [32].

0 -100 -200

PM pilot tone (2 GHz) AM pilot tone (8 GHz) calculation

-300 -400

-6

-4

-2

0

2

4

6

8

10

12

Input Power (dBm)

(a)

PM pilot tone (2 GHz) AM pilot tone (8 GHz) calculation

600 400 200 0 -200

0

10

20

30

40

50

DGD (ps)

(b)

Fig. 16. Effect of SPM and PMD on pilot-tone-based CD monitoring technique; monitoring Fig. 16 errors due to (a) SPM, (b) PMD.

CD monitoring techniques based on both AM and PM pilot tones could be affected by PMD, since the magnitude of pilot tone decreases as PMD increases [18]. We measured the effect of PMD on these techniques after replacing the SMF with a PMD emulator in Fig. 15(a). Thus, the total CD was 0 ps/nm. Figure 16(b) shows the maximum PMD-induced errors measured while varying the differential group delay (DGD) of PMD emulator. When the DGD was 30 ps, the maximum monitoring error was measured to be 420 ps/nm for the technique using AM pilot tones, while it was less than 30 ps/nm for the technique using PM pilot tones. Thus, we concluded that the monitoring technique using PM pilot tones provides superior performance to that of the technique using AM pilot tones by suppressing the effects of both SPM and PMD. 3.B.

Polarization-Mode-Dispersion Monitoring

Owing to the random birefringence of optical fiber, PMD has statistical and time-varying characteristics. Thus, an accurate PMD monitoring technique is vital for the efficient compensation of PMD. Recently, it has been proposed to monitor PMD by use of pilot tones [15]. However, as discussed in the previous section, this technique is sensitive not only to PMD, but also to CD. In this Subsection, we describe the operating principle of the PMD monitoring technique based on pilot tones and discuss its performance in the presence of CD. We then propose a novel technique using SSB pilot tones to mitigate this problem. In addition, we introduce a simple technique capable of monitoring CD and PMD simultaneously, without affecting each other, by polarization scrambling the optical signal [20].


526

3.B.1.

Principles

When the optical signal is transmitted through the fiber with nonnegligible PMD, the detected magnitude of pilot tone, P, can be described as [20] P ∝ 1 − 4γ (1 − γ) sin2 (π f ∆τ) ,

(3)

where ∆τ and γ are DGD and power ratio between the fast and slow axes, respectively, and f represents the tone frequency. This equation indicates that the detected magnitude of pilot tone, P, decreases as PMD increases. Thus, we could monitor the PMD of WDM signal by measuring the magnitude of pilot tone. For the use of pilot tones in the PMD monitoring technique, we should first determine the proper tone frequency. Figure 17 shows the magnitudes of pilot tones measured while varying the tone frequency and DGD. For this measurement, we used the same experimental setup shown in Fig. 15(a), after replacing the SMF with a PMD emulator. The state of polarization (SOP) of the signal incident on the PMD emulator was adjusted to minimize the magnitude of the detected pilot tone (i.e., γ = 0.5). The result shows that the resolution of this technique [i.e., ∆P/∆ (DGD)] could be improved by increasing the tone frequency, although it would reduce the measurement range. On the other hand, when DGD was larger than 50 ps, the power penalty (@BER = 10−9 ) was measured to be greater than 5 dB for the 10-Gbit/s NRZ signal (pattern length = 231 − 1). Thus, we concluded that the proper tone frequency would be 10 GHz if the maximum DGD were smaller than 50 ps. Under this condition, the MI of the pilot tone should be set to less than 14% to maintain the tone-induced power penalty within 0.5 dB [see Fig. 3(b)].

Measured Pilot Tone (a.u.)

1.2 1.0

DGD=0 ps

0.8

DGD=20 ps

0.6 DGD=30 ps 0.4 calculation measurement

0.2

DGD=40 ps DGD=50 ps

0.0

0

2

4

6

8

10

12

14

Tone Frequency (GHz) Fig. 17. Magnitude of pilot toneFig. versus 17 tone frequency and DGD.

The accuracy of the PMD monitoring technique based on the high-frequency pilot tone could be deteriorated by CD. This is because the phase difference between the lower and upper sidebands of a pilot tone could be changed by CD, which, in turn, decreases the magnitude of pilot tone as CD increases. This problem could be solved by using a SSB pilot tone to monitor PMD. The SSB pilot tone is inherently insensitive to CD, since it has only one sideband. Figure 18 shows the magnitudes of SSB pilot tones (tone frequency = 10 GHz) measured while varying both CD and DGD, in comparison with the conventional double sideband (DSB) pilot tones. As expected, the magnitude of the DSB pilot tone decreased as CD increased. However, by using the SSB pilot tone, we could substantially reduce this sensitivity to CD and monitor PMD with a reasonable accuracy even when CD was as high as 680 ps/nm.


527

Measured Pilot Tone (dB)

Measured Pilot Tone (dB)

5 0 -5 -10

CD=0 ps/nm CD=170 ps/nm CD=340 ps/nm CD= 680 ps/nm

-15 -20 -25 -30 -35

0

10

20

30

40

50

5 0 -5 -10

CD=0 ps/nm CD=170 ps/nm CD=340 ps/nm CD= 680 ps/nm

-15 -20 -25 -30 -35

0

10

DGD (ps)

20

30

40

50

DGD (ps)

(a)

(b)

Fig. 18. Magnitudes of pilot tones measured Fig. 18 while varying both DGD and CD: (a) when DSB pilot tone is used, (b) when SSB pilot tone is used.

3.B.2.

Polarization Scrambling Technique for Monitoring Both PMD and CD

As discussed in Subsection 3.B.1, the pilot-tone-based monitoring technique is sensitive to both CD and PMD. Thus, it is difficult to distinguish the effects of one from another. To overcome this problem, we proposed a simple technique to monitor both PMD and CD simultaneously by polarization scrambling the optical signal [35]. Figure 19 shows the experimental setup used for the demonstration of the proposed technique. We used a DFB laser operating at 1550 nm. The output power was set to be 3 dBm. We modulated this laser at 20 Gbit/s (pattern length, 231 − 1) by using a LiNbO3 modulator. The modulated signal was polarization-scrambled with use of rotating wave plates [33]. This process rotated the SOP of the optical signal periodically at the scrambling frequency. The scrambled signal was first sent to the PMD emulator to simulate the first-order PMD. The output of the PMD emulator traversed through SMF 0–40 km and split into two input ports of the monitoring module. One photodetector (PD1) was used for the detection of high-frequency component, while the other photodetector (PD2) was used to measure the total optical power. The output of PD1 was then filtered with a bandpass filter (center frequency, 9.814 GHz; passband, 300 MHz), and sent to the electrical power detector to measure the electrical power of the frequency component at 9.814 GHz (from 20-Gbit/s NRZ signal). The detected electrical power was sampled by use of an A/D converter and sent to the microprocessor to determine the first-order PMD and CD. Although we did not use pilot tone in this experiment, a highfrequency pilot tone (operating at 9.814 GHz) could be used to improve the monitoring sensitivity.

LD

IM PRBS

20 Gb/s

Polarization scrambler

PMD emulator

SMF 0~40 km

PD1 BPF

RF detector

fc=9.814 GHz

A/D converter

Microprocessor

PD2 Monitoring module

Fig.technique 19 Fig. 19. PMD and CD monitoring based on polarization scrambling.


528

Normalized Electrical Power (a.u.)

1.2 1.0 0.8 0.6 0.4 0.2 0.0

0

100

200

300

400

500

600

700

800

Chromatic Dispersion (ps/nm) Fig. 20. Electrical power of the frequency component at 9.814 GHz measured as a function Fig. 20 of CD.

We assumed that there is no higher-order PMD. Thus, the detected electrical power of the frequency component at f can be described as P (γ, ∆τ, D, L, f ) = 1 − 4γ (1 − γ) sin2 (π f ∆τ) × Pc (D, L, f ) , (4) where D and L are the dispersion coefficient and length of optical fiber, respectively. In this equation, we introduce the parameter, Pc (D, L, f ), to describe the effect of CD. To evaluate the effect of CD, we measured the electrical power of the frequency component at 9.814 GHz as a function of CD. During this measurement, we changed the length of SMF from 0 to 40 km (to vary CD), while DGD of the PMD emulator was set to be 0 ps. Thus, this result would reflect only the effect of CD, since the fiber used in this experiment had negligible PMD (< 0.1 ps/km1/2 ). Figure 20 shows that the measured electrical power decreased as we increased the fiber length (i.e., increased CD) [18]. However, Eq. (4) indicates that this power would also be dependent on PMD. Thus, for the accurate monitoring of CD, it would be necessary to eliminate the effect of PMD. This could be achieved with the transmission technique, which is based on the principal state of polarization (PSP) [34]. For example, if γ is either 0 or 1, the effect of PMD could be completely eliminated. In this case, Eq. (4) could be simplified as Pmax = Pc (D, L, f ) . (5) Thus we could monitor CD, without the deleterious effect of PMD, by measuring the electrical power when γ is either 0 or 1. On the other hand, the effect of PMD could be maximized when γ is 0.5. Equation (4) could then be expressed as Pmin = cos2 (π f ∆τ) × Pc (D, L, f ) .

(6)

By using Eqs. (5) and (6), we could estimate ∆τ, without the effect of CD, by using the following equation: ∆τ = cos−1 (2Pmin /Pmax − 1) /2π f . (7) Since we scrambled the polarization of optical signal at the transmitter, the measured optical power varied randomly (as γ was fluctuated between 0–1), as shown in Fig. 21. However, as expected in Eq. (7), the ratio between the maximum and minimum values was increased with PMD. Thus, we could monitor both CD and PMD simultaneously, without


529

Electrical Power (a.u.)

0.6 0.5

∆τ = 10 ps

0.4 0.3

∆τ = 30 ps

0.2 0.1 ∆τ = 50 ps 0.0 0.0

0.5

1.0

1.5

2.0

Time (sec) Fig. 21. Electrical power of the frequency component (9.814 GHz) measured when the Fig. 21 polarization of optical signal was scrambled.

450

50

400

∆τ = 40 ps

± 2.5 ps 30

∆τ = 30 ps 20

10

∆τ = 20 ps

0

2

4

6

Time (min)

(a)

8

10

DL=345 ps/nm

350

CD (ps/nm)

PMD (ps)

40

± 30 ps

300

DL=246 ps/nm

250 200

DL=163 ps/nm

150 100 50

0

2

4

6

8

10

12

14

Time (min)

(b)

Fig. 22. (a) Measured PMD while Fig. varying 22 CD (CD = 0, 83, 163, 246, 345 ps/nm), (b) measured CD while varying PMD (∆τ = 0, 10, 20, 30, 40, 50 ps).


530

the deleterious effect from one to another, by using Eqs. (5) and (7). Using the proposed technique, we measured PMD while varying both DGD (∆τ = 20, 30, and 40 ps) and CD (0, 83, 163, 246, and 345 ps/nm). Figure 22(a) shows that we could monitor the first-order PMD with accuracy better than ±2.5 ps, even when CD was varied from 0 to 345 ps/nm. This inaccuracy was mostly attributed to the incomplete polarization scrambling and polarization dependent loss (PDL) [35]. Figure 22(b) shows that we could also monitor CD with accuracy better than ±30 ps/nm, even in the existence of substantial PMD (∆τ = 0–50 ps). 4.

Summary

For the proper operation and management of a dynamic WDM network, it is essential to monitor various parameters such as optical power, wavelength, optical signal-to-noise ratio (OSNR), optical path, and so on directly in the optical layer. Pilot tones are well suited for this application since they are bound to follow their corresponding optical signals anywhere in the network. In addition, the pilot-tone-based technique is extremely cost-effective since it could monitor these parameters without using the expensive tunable optical filters and diffraction gratings. Accordingly, there have been many proposals to use pilot tones for monitoring various optical parameters in a WDM network. For example, it has been reported that the pilot tones could be used for monitoring the optical power, wavelength, OSNR, optical path, cross talk, and so on. In this paper, we reviewed these techniques and discussed their potential problems with possible solutions. Although the pilot-tone-based monitoring technique has many advantages, its performance could be seriously deteriorated by XGM and SRS. This would also limit the maximum network size that the pilot-tone-based monitoring could support. Thus, it appears that this technique would be most useful for metro network applications. For the use in the long-haul network with a large number of WDM channels, it would be necessary to restrict the number of channels to be monitored at a time (e.g., by using an optical bandpass filter). The pilot-tone-based monitoring technique could also be used for the adaptive compensators of CD and PMD. However, the main obstacle in these techniques is the difficulty in separating the effects of CD and PMD. To overcome this problem, we proposed using PM pilot tones for monitoring CD. This technique could provide superior performance to the technique using AM pilot tones by suppressing the effects of SPM and PMD. For the PMD monitoring technique, we demonstrated that the tolerance to CD could be substantially improved by using SSB pilot tones. In addition, we showed that both CD and PMD could be monitored simultaneously, without influencing one from another, by polarization scrambling the optical signal. Acknowledgments

We express our gratitude to our former students, C. J. Youn, J. K. Kim, H. Kim, D. K. Jeong, C. H. Kim, and S. K. Shin, for their contribution to this study. This research was supported in part by the NRL program of the Ministry of Science and Technology.

References and Links [1] P. A. Bonenfant, “Tutorial on protection and restoration in optical networks,” in Optical Fiber Communication Conference (OFC 1999) (Optical Society of America, Washington, D.C., 1999), FF1, pp. 104. [2] S. K. Shin, C.-H. Lee, and Y. C. Chung, “A novel frequency and power monitoring method for WDM network,” in Optical Fiber Communication Conference (OFC 1998) (Optical Society of America, Washington, D.C., 1998), WJ7, pp. 168–170.


531

[3] K. Otsuka, T. Maki, Y. Sampei, Y. Tachikawa, N. Fukushima, and T. Chikama, “A highperformance optical spectrum monitor with high-speed measuring time for WDM optical networks,” in European Conference on Optical Communications (ECOC 1997) (Institution of Electrical Engineers, London, 1997), 2, pp. 147–150. [4] M. Teshima, M. Koga, and K. Sato, “Performance of multiwavelength simultaneous monitoring circuit employing arrayed-waveguide grating,” IEEE J. Lightwave Technol. 14, 2277–2285 (1996). [5] K. J. Park, S. K. Shin, and Y. C. Chung, “Simple monitoring technique for WDM networks,” IEE Elec. Lett. 35, 415–417 (1999). [6] C. J. Youn, S. K. Shin, K. J. Park, and Y. C. Chung, “Optical frequency monitoring technique using arrayed-waveguide grating and pilot tones,” IEE Elec. Lett. 37, 1032–1033 (2001). [7] G. Rossi, T. E. Dimmick, and D. J. Blumenthal, “Optical performance monitoring in reconfigurable WDM optical networks using subcarrier multiplexing,” IEEE J. Lightwave Technol. 18, 1639–1648 (2000). [8] G. R. Hill, P. J. Chidgey, F. Kaufhold, T. Lynch, O. Sahlen, M. Gustavsson, B. Lagerstrom, G. Grasso, F. Meli, S. Johansson, J. Ingers, L. Fernandez, S. Rotolo, A. Antonielli, S. Tebaldini, E. Vezzoni, R. Caddedu, N. Caponio, F. Testa, A. Scavennec, M. J. O’Mahony, J. Zhou, A. Yu, W. Sohler, U. Rust, and H. Herrmann, “A transport network layer based on optical network elements,” IEEE J. Lightwave Technol. 11, 667–678 (1993). [9] F. Heismann, M. T. Fatehi, S. K. Korotky, and J. J. Veselka, “Signal tracking and performance monitoring in multi-wavelength optical network,” in European Conference on Optical Communication (ECOC 1996) (Institution of Electrical Engineers, London, 1996), WeB.2.2, pp. 47–50. [10] H. C. Ji, K. J. Park, J. K. Kim, S. K. Shin, M. J. Jang, and Y. C. Chung, “Demonstration of optical path monitoring technique using dual tones in all-optical WDM transport network,” in Proceedings of the OptoElectronics and Communication Conference (OECC 2000) (IEEE, 2000), 14A4-4, pp. 476–477. [11] H. C. Ji, J. K. Kim, and Y. C. Chung, “Optical path and crosstalk monitoring technique using pilot tones in all-optical WDM transport network,” in Asia-Pacific Optical and Wireless Communications Conference (APOC 2001) (SPIE, 2001), 4584-2, pp. 28–33. [12] M. Murakami, T. Imai, and M. Aoyama, “A remote supervisory system based on subcarrier overmodulation for submarine optical amplifier systems,” IEEE J. Lightwave Technol. 14, 671– 677 (1996). [13] H. S. Chung, S. K. Shin, K. J. Park, H. G. Woo, and Y. C. Chung, “Effects of stimulated Raman scattering on pilot-tone-based supervisory technique,” IEEE Photonics Technol. Lett. 12, 731– 733 (2000). [14] T. E. Dimmick, G. Rossi, and D. J. Blumenthal, “Optical dispersion monitoring technique using double sideband subcarriers,” IEEE Photonics Technol. Lett. 12, 900–902 (2000). [15] S. M. R. Motaghian Nezam, Y. Wang, M. Hauer, S. Lee, and A. E. Willner, “Simultaneous PMD monitoring of several WDM channels using subcarrier tones,” in Conference on Lasers and Electro-Optics (CLEO 2001), Vol. 56 of OSA Trends in Optics and Photonics Series (Optical Society of America, Washington, D.C., 2001), CFE1, pp. 561–562. [16] M. N. Petersen, Z. Pan, S. Lee, S. A. Havstad, and A. E. Willner, “Online chromatic dispersion monitoring and compensation using a single inband subcarrier tone,” IEEE Photonics Technol. Lett. 14, 570–572 (2002). [17] K. J. Park, C. J. Youn, J. H. Lee, and Y. C. Chung, “Chromatic dispersion monitoring technique in WDM network,” in Optical Fiber Communication Conference (OFC 2002) (Optical Society of America, Washington, D.C., 2002), ThGG88, pp. 735–737. [18] K. J. Park, C. J. Youn, J. H. Lee, and Y. C. Chung, “Performance comparisons of chromatic dispersion-monitoring techniques using pilot tones,” IEEE Photonics Technol. Lett. 15, 873– 875 (2003). [19] C. J. Youn, “Optical performance monitoring techniques for WDM networks,” Ph.D. dissertation [Korea Advanced Institute of Science and Technology (KAIST), 2003]. [20] K. J. Park, H. Kim, J. H. Lee, C. J. Youn, S. K. Shin, and Y. C. Chung, “Polarization-mode dispersion monitoring technique based on polarization scrambling,” IEE Elec. Lett. 38, 83–85 (2002).


532

[21] ITU-T Study Group 15, “Tone modulation for suppressing stimulated Brillouin scattering and for channel identification on systems using in-line OFAs and WDM,” Contribution, Question 25/15 (1994). [22] Y. Sun, A. A. M. Saleh, J. L. Zyskind, D. L. Wilson, A. K. Srivastava, and J. W. Sulhoff, “Time dependent perturbation theory and tones in cascaded erbium-doped fiber amplifier systems,” IEEE J. Lightwave Technol. 15, 1083–1087 (1997). [23] A. K. Srivastava, J. L. Zyskind, Y. Sun, J. Ellson, G. Newsome, R. W. Tkach, A. R. Chraplyvy, J. W. Sulhoff, T. A. Strasser, C. Wolf, and J. R. Pedrazzani, “Fast-link control protection of surviving channels in multiwavelength optical networks,” IEEE Photonics Technol. Lett. 9, 1667– 1669 (1997). [24] J. H. Jang, S. K. Shin, H. Kim, K. S. Lee, K. J. Park, and Y. C. Chung, “A cold-start WDM system using a synchronized etalon filter,” IEEE Photonics Technol. Lett. 9, 383–385 (1997). [25] K. J. Park, S. K. Shin, H. C. Ji, H. G. Woo, and Y. C. Chung, “A multi-wavelength locker for WDM system,” in Optical Fiber Communication Conference (OFC 2000) (Optical Society of America, Washington, D.C., 2000), WE4, pp. 73–75. [26] E. Kong, F. Tong, K.-P. Ho, L.-K. Chen, and C.-K. Chan, “An optical-path supervisory crossconnects using pilot-tones,” in Conference on Lasers and Electro-Optics (CLEO 1999) (Optical Society of America, Washington, D.C., 1999), FT4, pp. 1281–1282. [27] C.-K. Chan, E. Kong, F. Tong, and L.-L. Chen, “A novel optical-path supervisory scheme for optical cross connects in all-optical transport network,” IEEE Photonics Technol. Lett. 10, 899901 (1998). [28] K. U. Chu, C. H. Lee, and S. Y. Shin, “Optical path monitoring based on the identification of optical cross-connect input ports,” in Optical Fiber Communication Conference (OFC 2000) (Optical Society of America, Washington, D.C., 2000), FJ5, pp. 158–160. [29] J. K. Kim, H. C. Ji, H. S. Chung, C. H. Kim, S. K. Shin, D. H. Hyun, and Y. C. Chung, “Demonstration of fast restorable all-optical WDM network,” IEICE Trans. Commun. E 84-B, 1119– 1126 (2001). [30] C. D. Poole, R. W. Tkach, A. R. Chraplyvy, and D. A. Fishman, “Fading in lightwave systems due to polarization-mode dispersion,” IEEE Photonics Technol. Lett. 3, 68–70 (1991). [31] T. Kato, Y. Koyano, and M. Nishimura, “Temperature dependence of chromatic dispersion in various types of optical fibers,” in Optical Fiber Communication Conference (OFC 2000) (Optical Society of America, Washington, D.C., 2000), TuG7, pp. 104–106. [32] K. J. Park, C. J. Youn, J. H. Lee, and Y. C. Chung, “Effect of self-phase modulation on groupvelocity dispersion measurement technique using PM-AM conversion,” IEE Electron. Lett. 38, 1247–1248 (2002). [33] J. H. Lee, D. K. Jung, C. H. Kim, and Y. C. Chung, “OSNR monitoring technique using polarization-nulling method,” IEEE Photonics Technol. Lett. 13, 88–90 (2001). [34] T. Ono, Y. Yano, L. D. Garrett, J. A. Nagel, M. J. Dickerson, and M. Cvijetic, “10 Gbit/s PMD compensation field experiment over 452 km using principle state transmission method,” in Optical Fiber Communication Conference (OFC 2000) (Optical Society of America, Washington, D.C., 2000), paper PD44. [35] K. J. Park, J. H. Lee, C. J. Youn, and Y. C. Chung, “A simultaneous monitoring technique for polarization-mode dispersion and group-velocity dispersion,” in Optical Fiber Communication Conference (OFC 2002) (Optical Society of America, Washington, D.C., 2002), WE4, pp. 199– 200.


533