Experimental and numerical investigation of bit-wise phase-control OTDM transmission

Experimental and numerical investigation of bit-wise phase-control OTDM transmission

Sergejs Makovejs,* Giancarlo Gavioli, Vitaly Mikhailov, Robert I. Killey and Polina Bayvel

Optical Networks Group, Department of Electronic and Electrical Engineering, University College London, Torrington Place, London WC1E 7JE, United Kingdom

*Corresponding author: [email protected]

Abstract: The impact of varying the phase relationship between adjacent OTDM channels is investigated in 80 Gbit/s transmission experimentally and numerically. A fiber-based coherent multiplexer is proposed for OTDM experiments - a phase shifter in the multiplexer and an external phase control circuit are used to set and maintain the phase difference. It is demonstrated that the optimum modulation format for maximum transmission distance strongly depends on pulse width, e.g. 120º-RZ provides the best performance for pulse width of 8 ps; however, 90º-RZ is advantageous when pulse width is reduced to 2 ps. Power in ‘zero’ bit slots and amplitude jitter are calculated to demonstrate that the performance variation is due to intra-channel four-wave mixing (IFWM) and different receiver sensitivity at back-to-back. We also show that phase modulation formats are sensitive to optical filtering.

©2008 Optical Society of America

OCIS codes: (060.2330) Fiber optics communications; (060.5060) Phase modulation.

References and links

1. H. G. Weber, S. Ferber, M. Kroh, C. Schmidt-Langhorst, R. Ludwig, V. Marembert, C. Boerner, F. Futami, S. Watanabe, and C. Shubert, “Single channel 1.28 Tbit/s and 2.56 Tbit/s DQPSK transmission,” Electron. Lett. 42, 178-179 (2006).

2. S. Weisser, S. Ferber, L. Raddatz, R. Ludwig, A. Benz, C. Boerner, and H. Weber, “Single- and alternating-polarization 170-Gb/s transmission up 4000 km using dispersion-managed fiber and all-Raman amplification,” IEEE Photon. Technol. Lett., 18, 1320-1322, (2006).

3. 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).

4. L. Möller, Y. Su, X. Liu, J. Leuthold, and C. Xie, “Generation of 160 Gb/s carrier-suppressed RZ signals,” in Proc. 29th European Conf. on Opt. Commun. (ECOC2003), Mo3.6.3, 2003.

5. A. Hirano, M. Asobe, K. Sato, K. Yonenaga, Yutaka Miyamoto, H. Takara, I. Shake, H. Miyazawa, and M. Abe, “Dispersion tolerant 80 Gbit/s optical-time-division multiplexing using a duty- and phase-control technique,” in Proc. 25th European Conf. on Opt. Commun. (ECOC1999), We C1, 1999.

6. S. Pitois, “Optimisation of phase alternation in 160 Gbit/s transmission systems,” Opt. Commun. 242, 457-461 (2004).

7. P. Johannisson, D. Anderson, M. Marklund, A. Berntson, M. Forzati, and J. Martensson, “Suppression of nonlinear effects by phase alternation in strongly dispersion-managed optical transmission,” Opt. Lett. 27, 1073-1075 (2002).

8. S. Appathurai, V. Mikhailov, R. Killey, and P. Bayvel, “Investigation of the optimum alternate-phase RZ modulation format and its effectiveness in the suppression of intrachannel nonlinear distortion in 40-Gbit/s transmission over standard single-mode fiber,” IEEE J. Sel. Top. Quantum Electron. 10, 239-249 (2004).

9. I. Morita and N. Edagawa, “Study on optimum OTDM signals for long-distance 40 Gbit/s transmission,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2002), TuA4, 2002

10. D. Gill, A. Gnauck, and X. Liu, X. Wei, and Y. Su, “π/2 alternate-phase on-off keyed 42.7 Gb/s long-haul transmission over 1980 km of standard single-mode fiber,” IEEE Photon. Technol. Lett. 16, 906-908 (2004).

11. S. Randel, B. Konrad, A. Hodžić, and K. Petermann, “Influence of bitwise phase changes on the performance of 160 Gbit/s transmission systems,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2002), P3.31, 2002.

(C) 2008 OSA 10 November 2008 / Vol. 16, No. 23 / OPTICS EXPRESS 18725#100847 - $15.00 USD Received 29 Aug 2008; revised 18 Oct 2008; accepted 27 Oct 2008; published 29 Oct 2008

12. L. Boivin and A. Chraplyvy, “Testing optical time-division multiplexed transmission systems with interleaved bit sequences,” in Tech. Dig. Optical Fiber Communications Conf. (OFC2002), WM35, 2000.

13. A. Ellis, PhD Thesis, 1997. 14. P. Winzer, M. Phenningbauer, M. Strasser, and W. Leeb, “Optimum filter bandwidths for optically

preamplified NRZ receivers,” J. Lightwave Technol. 19, 1263-1273, (2001). 15. C. Laperle, B. Villeneuve, Z. Zhuhong, D. McGhan, H. Sun, and M. O’Sullivan, “WDM performance and

PMD tolerance of a coherent 40-Gbit/s dual-polarization QPSK transceiver,” J. Lightwave Technol. 26, 168-175 (2008).

16. D. Fonseca, A. Cartaxo, and P. Monteiro, “Influence of the extinction ratio on the intrachannel nonlinear distortion of 40-Gb/s RZ transmission systems over standard fiber,” J. Lightwave Technol. 25, 1447-1457 (2007).

1. Introduction

Optical time-division multiplexing (OTDM) is an attractive technique to investigate the performance of high-speed transmission systems [1, 2]. In a typical laboratory OTDM experiment, signal replicas are displaced in time, with respect to each other, in a fiber-based multiplexer, and then recombined to obtain a high-rate data stream. Ambient temperature variations in the multiplexer and mechanical vibrations cause random phase fluctuations and, therefore, advanced modulation formats based on fixed phase relationships between adjacent OTDM channels cannot be generated. However, it has been shown in 40 Gbit/s experiments, that phase modulation formats, such as, CS-RZ and π/2-RZ, can be more tolerant to linear and nonlinear impairments, allowing longer transmission distances. To date, bit-wise (or, coherent) π/2-RZ and CS-RZ OTDM signals have been generated (and transmitted) by either varying the operation temperatures of electro-absorption modulators (EAMs) in an OTDM free-space multiplexer [3], controlling the polarization state of a pump signal in a highly nonlinear fiber [4] or using a PLC-type OTDM multiplexer with a phase controller [5].

However, the optimum phase difference between the adjacent channels, and thus the modulation format, still remains a question of debate in transmission regime where significant pulse spreading occurs, i.e. 40 Gbit/s and higher. In fact, it has been mathematically shown that the maximum suppression of ghost pulses due to intra-channel four-wave mixing (IFWM) is achieved using π/2 phase difference between adjacent pulses [6, 7]. It was also suggested in [8] that amplitude distortion due to phase-to-intensity conversion is higher in the case of π/2 phase modulation, and the overall performance is better in case of π [8, 9]; however, this contradicts the experiments in [10] where more than a 2 dB improvement in Q-factor was shown for π/2 compared with π. At higher bit rates, in [6, 11] 160 Gbit/s OTDM Q-factors and eye opening penalties have been simulated as a function of adjacent OTDM channels phase difference, and an optimum performance was found in case of π/2-RZ. Finally, in [5] simulations showed that the optimum phase difference strongly depends on the duty cycle.

In this paper we experimentally investigate the impact of a wide and continuously variable range of phase values between adjacent OTDM channels on the transmission performance at 80 Gbit/s. This is achieved with a novel fiber-based coherent OTDM multiplexer, which allows for bit-wise phase control. We used it to investigate modulation formats with different phase relationships between the adjacent OTDM channels in a recirculating loop transmission experiment at 80 Gbit/s and compared with numerically obtained results.

2. Experimental set-up

Figure 1 shows the experimental setup. A 40 Gbit/s RZ signal was generated by modulating the output of an external cavity laser (ECL) at 1545 nm with two EAMs: one for data encoding (27-1 PRBS was used due to limitations in clock recovery operation with longer patterns) and one for pulse carving, generating RZ pulses with 8 ps FWHM, 38 dB OSNR and 15 dB extinction ratio. The optical signal was launched into a fiber-based coherent OTDM multiplexer to generate an 80 Gbit/s signal. The multiplexer was designed using


PC 50/5050/50

FEEDBACK CIRCUIT

OPTICAL RECEIVER

40G PPG

SMFDCF

LOOP AOM

EDFA

TX AOM

ECL EAMEAM

40G DATA 40G CLOCK

EAM DEMUX

CRU

40GHz

40G BERT

φ

ATT

t-τ

OF

Fig. 1. Experimental set-up for 80 Gbit/s transmission.

polarization-maintaining (PM) fiber-coupled components to generate co-polarized OTDM tributaries; an additional 5 m of PM fiber was added for decorrelation at long bit patterns [12].

The multiplexer path length difference was less than the laser coherence length of ~45 m, which allowed for coherent interference when two delayed signals were recombined [13]. The generated pulses (8 ps FWHM) partially overlap in the two multiplexer arms, even though they have a relative time-shift. The degree of overlap and an interference product, determined by the relative phase shift, is used as an input signal to the feedback circuit (an electrical integrator with 1 kHz bandwidth) used to set and track the voltage to the phase shifter to generate the stable phase difference between the adjacent OTDM channels. Variations in temperature faster than 1 kHz were minimized by placing the multiplexer in a temperature-controlled environment. Alternatively, a modulator bias controller (ditherless) can be used as a feedback circuit. In the experiment we used the same laser for both the feedback circuit and data transmission to avoid the need for multiple laser wavelength stabilization. The set-up needs a slight modification in the case of no pulse overlap (e.g. 2 ps pulses). It involves using a ECL second CW output as a reference signal to be counter-propagated in the multiplexer with the data (to avoid extinction ratio degradation) and detected by the feedback circuit.

The resultant 80 Gbit/s signal was then launched into a recirculating loop, consisting of a 100 km SMF span, fully dispersion compensated at 1545 nm and flat-top optical filter (OF) with 100 GHz 0.5 dB bandwidth. The noise figure of the loop EDFAs was 4.5 dB. After transmission, a 40 GHz clock signal was extracted using an EO-PLL clock recovery unit (CRU), and used to drive an EAM-based optical time-demultiplexer. The demultiplexed 40 Gbit/s signal was detected by a 40 Gbit/s optical receiver, after which the BER was counted. The optical filter used in the receiver was the same as the one in the recirculating loop.

3. Experimental and numerical results

The variation of the phase shift between OTDM channels introduces a power change in the carrier and the 40 GHz clock components. Figure 2 shows the ratio between the carrier and 40 GHz spectral components (clock-carrier power difference, CCPD) calculated numerically as a function of phase shift between OTDM channels. This result leads to two considerations; first, experimentally measured CCPD can be used to set the desired phase shift values in coherent OTDM multiplexer. Second, different phase modulated OTDM formats will have different optimum optical filter bandwidth due to associated spectral reshaping [14]. Phase values corresponding to CCPD = -14, -12, -8, -4, 0 and 4 dB were set during the experiment. Pure RZ


and CS-RZ were outside the operating range of the feedback circuit and, therefore, were not implemented during an experiment.

In back-to-back 80 Gbit/s measurements, a 4 dB improvement in receiver sensitivity at BER of 10-9 was measured when increasing the phase difference from 32°-RZ (CCPD = -14 dB) towards 135°-RZ (CCPD = 4 dB). We attribute the improvement in the receiver sensitivity to reduced inter-symbol interference (ISI) between the adjacent pulses modulated with orthogonal phase values. Simulations showed that 4 dB back-to-back penalty disappears when shorter pulses are used (e.g. 2ps), due to minimized pulse overlap, and hence, ISI.

1544.8 1544.9 1545 1545.1 1545.2 1545.3 1545.4 1545.5

-25

-20

-15

-10

-5

Wavelength (nm)

Am

plitu

de (

dBm

)

-N dBm

(b)

0 20 40 60 80 100 120 140 160 180

-25

-20

-15

-10

-5

0

5

10

15

20

CC

PD

(d

B)

Phase shift (degrees)

(a)

1544.7 1544.8 1544.9 1545 1545.1 1545.2 1545.3 1545.4 1545.5 1545.6-30

-25

-20

-15

-10

Wavelength (nm)A

mpl

itude

(dB

m) + N dBm

(c)

Fig. 2. (a) Clock-carrier power difference (CCPD) dependance on phase shift between adjacent OTDM channels. Inset: eye-diagrams for 32°-RZ (CCPD = -14 dB) and 135°-RZ (CCPD = 4 dB) at the transmitter (b) Measured OTDM spectra at the transmitter with (b) clock component weaker than carrier by N dB. (c) clock component stronger than carrier by N dB (both over 0.01 nm bandwidth).

-20 0 20 40 60 80 100 120 140 160 180 200

600

700

800

900

1000

1100

1200

1300

1400

1500

Tra

nsm

issi

on

dis

tan

ce (

km)


Experimentally Numerically (optimized filtering)

Fig. 3. Maximum transmission distance at BER<3.8*10-3.

Transmission experiments were carried out, and the maximum transmission distances for

BER < 3.8*10-3 [15] were measured as a function of the phase difference between adjacent OTDM channels. The launch power into the SMF was optimized for each modulation format and was in the range of -1 to 1.5 dBm; constant launch power into the DCF (0 dBm) was maintained during the experiment. Figure 3 shows the comparison of experimental and numerical results, obtained by simulating the link used in the experiment. A flat-top 2nd-order Gaussian filter with variable bandwidth was used in numerical simulations to enhance the transmission performance. Both results show a similar trend in terms of the phase shift; however, optimizing the filter shape and bandwidth yielded an increased transmission distance, compared with experimental results where fixed-bandwidth filter was used.

32°-RZ

135°-RZ


A differential phase modulation of 120 degrees between adjacent pulses (120º-RZ) yielded the longest transmission, which we attribute to a combination of two factors: its increased tolerance to IFWM and increased receiver sensitivity (comparing with RZ or 90°-RZ) due to decreased ISI. In fact, IFWM is a phase sensitive effect determining the strength of the ghost pulses and amplitude jitter effects between interacting channels. A limited extinction ratio (15 dB) also impacts the modulation format tolerance to IFWM, as residual power in ‘zeros’ will also contribute to IFWM generation dynamics [16]. This is especially important when an EAM, with inherently limited extinction ratio, is used as a pulse carver. It should be noted that transmission was limited by nonlinearity both in SMF and DCF.

Figure 4 shows power in ‘zero’ bit slots (combines ghost pulse power, limited extinction ratio and power due to pulse overlap) and amplitude jitter of 2056 bits of a 231-1 PRBS pattern

after 100-km for 8 ps and 2 ps FWHM pulse width, calculated as a function of phase shift between adjacent channels. Noise from EDFA was neglected and a 13 dBm input launch power was used to investigate signal distortion only due to intra-channel nonlinearities. Both pulse power within ‘zeros’ and amplitude jitter in ‘ones’ were minimized for 120º-RZ and 8 ps pulses. We also found that for phase differences lower than 60º, the power in ‘zeros’ and amplitude jitter remained nearly constant; this is in good agreement with numerical results in Fig. 3, where low phase shift values yielded the same transmission distance. However, when pulse width was decreased to 2 ps, 90º-RZ resulted in maximum suppression of IFWM effects, with symmetrical curves for amplitude jitter and power in ‘zeros’. In the case of 8 ps pulses, however, initially overlapping pulses needed higher phase separation to cause destructive interference of the IFWM components [8]. In addition, ISI contributes to the total power in ‘zeros’, which results into a strong dependence of the power in ‘zero’ slots on the phase shift for wide pulses (8 ps). Eye diagrams for several phase formats are shown in Fig. 5.

-20 0 20 40 60 80 100 120 140 160 180 2000.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

0.24

Po

wer

in 'z

ero

s'


8 ps 2 ps

-20 0 20 40 60 80 100 120 140 160 180 200

0.260.280.300.320.340.360.380.400.420.440.460.480.500.520.540.560.58

Am

plit

ud

e jit

ter


8 ps 2 ps

(a) (b)

Fig. 4. (a) Ghost pulse power (b) and amplitude jitter (both normalized to signal peak power) as a function of phase shift between adjacent OTDM channels.

Fig. 5. Simulated eye-diagrams after 100 km propagation (13 dBm input power) for 2 ps (a-d) and 8 ps (e-h) pulses. (a) RZ (b) 90°-RZ (c) 120°-RZ (d) CS-RZ (e) RZ (f) 90°-RZ (g) 120°-RZ (h) CS-RZ.

To further investigate the impact of the phase modulation and the pulse width on the transmission distance at 80 Gbit/s, we carried out a set of simulations to obtain the maximum transmission distance as a function of both the applied phase shift and FWHM pulse width (Fig. 6). Previously the impact of duty cycle in RZ, π/2-RZ and CS-RZ has been investigated for only linear propagation regime with respect to dispersion tolerance [5]. Our simulations

(a) (b) (c) (d) (e) (f) (g) (h)


show that the π/2-RZ modulation format is optimal with 2 ps pulses (in agreement with [11]); however, the optimum phase difference increases for larger pulse width (Fig. 6(a)). Therefore, we can conclude that the optimum phase relationship for maximum transmission distance depends on the OTDM duty cycle. This is due to an increase in receiver sensitivity for out-of-phase formats and the dependence of the phase shift on tolerance to IFWM when changing the pulse width. This consideration is particularly important as, at high bit rates, low duty cycles become infeasible with EAM (cascading many EAMs will inevitably lead to OSNR degradation due to additional EDFAs). Thus, it is important to assess the modulation formats performance with the pulse width that can be practically generated by EAMs.

To understand the effect of the optical filter bandwidth on performance of phase modulation formats, the variation in transmission distance as a function of pulse width and phase shift, when optimized optical filtering for each modulation format is applied, was investigated. In the simulations, a 2nd-order optical Gaussian filter bandwidth was varied from 130 - 200 GHz (FWHM) to ensure longest transmission distance for a particular modulation format, and the results are shown in Fig. 6(b). It is well-known that optimum filtering is a trade-off between reduction of detection noise and partial rejection of the spectral components [14]. We observed that RZ and CS-RZ typically required slightly lower bandwidth than π/2-RZ or 120º-RZ, which contain more spectral components; a narrow filter, therefore, partially removes these components and increases eye-opening penalty.

9.0E21.0E3

1.1E3

1.2E3

1.3E3

8.0E2

1.4E31.5E3

2 3 4 5 6 7 80

20

40

60

80

100

120

140

160

180

7.0E2

1.1E3

Ph

ase

shif

t (d

egre

es)

Pulse width (ps)

1.1E3

1.2E3

1.3E31.4E31.5E3

1.1E3

1.6E3

2 3 4 5 6 7 80

20

40

60

80

100

120

140

160

180

1.4E3

1.5E3

1E3

1.8E3

Ph

ase

shif

t (d

egre

es)

Pulse width (ps)

1.7E3

(a) (b)

Fig. 6. Simulated maximum transmission distance (km) at 80 Gbit/s as a function of pulse width and phase shift between OTDM channels. (a) Fixed-bandwidth filter as in an experiment. (b) Optimized filter.

4. Conclusions

The impact of different phase relationships between adjacent OTDM channels in a novel coherent multiplexer was investigated in a recirculating loop 80 Gbit/s experiment. Overall, the transmission distance is sensitive to the phase shift between OTDM channels, which can be explained by the fact that IFWM is a phase sensitive effect – different phase shifts between adjacent channels will produce ghost pulses with different powers and amplitude jitter. In 80 Gbit/s experiment over SMF fiber with 8 ps pulse width, the transmission distance at FEC limit reached 1000 km with 113º and 135º phase shifts. Simulations have also shown that the optimum phase shift depends on the pulse width – when decreased to 2 ps the maximum transmission distance was achieved for π/2 phase shift. We have also assessed the impact of optical filter bandwidth on achievable distance for different phase formats. This is the first experimental investigation on the influence of a continuously variable range of phase values between OTDM channels on the transmission performance at 80 Gbit/s.

Acknowledgments

The authors are grateful to Yokogawa Electric Corporation, EU IP NOBEL, the Leverhulme Trust and the Royal Society for financial support.


Experimental and numerical investigation of bit-wise phase-control OTDM transmission

Documents

Transcript of Experimental and numerical investigation of bit-wise phase-control OTDM transmission