[13]Influence of the dispersion map on limitations due to cross-phase modulation in wdm multispan transmission systems

7/27/2019 [13]Influence of the dispersion map on limitations due to cross-phase modulation in wdm multispan transmission s

1/3

MF4-1

Influence of the dispersion map on limitations due to CFOSS-phase modulation in WDM multispan transmisslion systems

C. Furst, C. Scheerer, G. Mohs, J.-P. Elbers, C. GlingenerSIEMENS AG, Advanced Transport Systems,ICN TR ON D T, Hofmannnstr.51, 81359 Munchen, GermanyTel. +49-89-722 44612, Fa x +49-89-722 45146, [email protected]

Abstract: We determine system limitations due to cross-phase m odulation(XPM) for non-zerodispersion shifted fiber by experiments and simulations. For the worst case of full dispersioncompensation in each span we find a maximum power law whereas for appropriateundercompensation XPM can be efficiently suppressed.02000 Optical Societyof AmericaOCIS codes: (060.2330) Fiber optics communications,(060.4370)Nonlinear optics,fibers

~ 1. Introduction

Cross-phase modulation (XPM ) has been identified as a limiting effect for the transmission reach in dense WD Msystems particularly on low dispersion fiber types as e.g. LEAF and TrueW ave. Various papers have been published

highlighting the influence of XPM in particular system configurations and their potentially detrimental impact on thesystem performance (e.g.[ 1,2]). Analytical models[3-51 have been developed and nu merical simulations[6 ] havebeen carried out for certain examples. Recently, a technique was published to suppress XPM distortions byintroducing time d elays between the different channels after each transmission span[7].

In this paper, we give an extended analysis of the system limitations imposed by XPM in dependence of the,dispersion map, employing both experiments and simulations. We focus on post-compensated systems and vary boththe inline and the total accumulated dispersion. In o ur experiments, uncorrelated channels are transmitted overseveral reamplified spans of non-zero dispersion shifted fiber in a recirculating loop. From the results we directlyderive design rules for the dispersion map for optimum XPM suppression. We show for the first time that amaximum power law holds for full inline dispersion compensation in multispan transmission which allows a fastestimate of the system limitations due to XPM .

2. Experimental setup and simulation model

For the first experiments, we investigate the copropagationof two fully independent wavelength channels with achannel spacing of 100 GHz. Both signals( N R Z ) are generated by sepa rate pattern generators (bothPRBS z7- l )with slightly different bit rates of 9.952 Gb/s for the reference channel and10.000 Gb/s fcir the distortin g channel.In this way we ensure all possible relative bit phases between the two signals to be included and avoid themeasurement of special cases due to specific launch conditions at the fiber input. The effect of XPM stronglydepends on the relative polarization of the copropagating signals. The worst case is obtained by launching thechannels with the same polarization state at the fiber input. The loop consists of an fiber amplifier followed by50 km of LEA F fiber (D=2.4 ps/nm/km) and a dispersion compensating module. The inline compensation for eachround trip is m odified by changing the m odule or inserting additional pieces of fiber. At the loop o utput, the signalpasses an additional fiber section where the total ac cumulated dispersion of the linkis controlled. The optical signal-to-noise ratio (OSNR)is adjusted with an optical attenuator before passing the EDFA pream plifier. The referencesignal is selected by a filter and detected by a receiver from our produ ct line allowing tlie measurement of bit errorrates

The deg radation of the reference signal due to XPM is analyzed by means of the OSN R required for a certain biterror rate, here chosen to be Signal distortions dueto fiber nonlinearity manifest themselves as a higherrequired OSNR with respect to linear transmission. We define the maxim um transmission distance as the span count(loop circulations) causing anOSNRpenalty of 1dB .

We also investigate the,XPM -induced performance degrad ation by numerical simulations using the split-stepFourier method. The simulation setup is chosen to match the experimental parameters as accurately as possible.Again, the channels c an y signals with slightly different bit rates. A large number of bits is calculated to include allpossible relative bit phases. The optical noise is included analytically at the receiver. For Ihe calculation of the biterror rate for a givenOSNR, all bit lines of the e ye diagram are taken into account.

7/27/2019 [13]Influence of the dispersion map on limitations due to cross-phase modulation in wdm multispan transmission s

2/3

MF4-2

' 3. Results and discussion

Firstly, we exam ine the evolution of theOSNR penalty with the span coun t for different inline compensation in eachspan for two interfering channels( Fig. 1). The total accum ulated dispersion of the link is kept con stant at630 p s h m

~ for all cases and span coun ts which ensures a sufficiently high penalty for measuremen t. The op tical launch power

total accumulated dispersion630 pdn m each case

19

total accumulated dispersion630 pdn m each case I

1 5 ' , , , , , , , , , ,0 1 2 3 4 5 6 7 8 9

4 spans , P,=BdBm, PxOdBm 1 Inline camp

I

0 -600 -400 -200 0 200 400 600 800 1000

Total Accumulated Dispersion (pdnm )

1

- 2 0 2 4 6 8 1 0 1 2

Distorting Ch annel Power P, (dBm)

Fig. 4 Span count N vs. input power Pd or experiments andsimulation. The product N Pd s shown to he a constant.

-200 -150 -100 -50 0 50 100 150 200

Number of Spans Intine Dispersion D l w pps/nm)

Fig. 1 Evolution of the required OSNR in 0.1 nm bandwidth

for different inline compensations. filled symbols:experiments, hollow: simulation (630 pdnm).

Fig. 2 Experimentally required OSNR after four transmitted

spans. The total accumulated dispersion is always the same

of the distorting channel is Pd=8 dBm . For the reference channel, we ch ose a pow er of onlyP,f=O dBm thuseffectively suppressing the influence of self-phase-modu lation and fo ur-wave-m ixing. For full inline comp ensation(Dloop=Osh m fo r one loop round trip), the required OSNR rapidly increases with the span count, whereas for ainline dispersion of 80 ps/nm the XPM degfadation is fully canceled after5 spans. These results are reproduced inour simu lations, shown by the hollow sym bols forDloop=O s h m a nd DIoO,=8O sh m inline dispersion.

This behavior is further highlighted in the experiments shown in Fig.2, where we transmit four spans atPd=6 dBm varying the inline dispersion but still maintaininga constant accum ulated dispersion. We clearly identifythe full compensation as the worst case explained by in-phase accumulation of the X PM distortions for subsequentspans. The best case of80 p sh m span dispersion corresponds to a time delay of64 ps between the two signals afterone round trip. The XPM distortions generated in the different spans partly add destructively which leadsto acancellation of the penalty in good agreemen t with ref.[7].

In Fig. 3, we check the influenceof the total accum ulated dispersion on the d ifferent inline comp ensationschemes transmitting four spans(Pd=8 dBm ). Compa red to the case of full inline com pensation, the dispersiontolerance is nearly doubled for the undercompensated scheme. The m inimumOSNR is the same for both schemesafter four spans. Further investigations show that the minim um OSNR is increasing for a higher span count in thecase of full inline compensation, but keep s constant for the undercom pensated scheme.

In the follow ing we concentrate on the worst case of full inline dispersion compensation. Fig.4 displays thespan count N with an OSN R penalty of 1dB for different channel powersP d . To decrease the quantization of the datapoints, non-integer span counts are interpolated from the m easuredOSNR penalties. The experiments show that the

20

-0F

d 18Wmb

c

a:2

16

U

3

2

P

.-

14I

Fig. 3 Dispersion tolerance for worst and best case inlinecompensation for 2-channel experiments.

7/27/2019 [13]Influence of the dispersion map on limitations due to cross-phase modulation in wdm multispan transmission s

3/3

MF4-3

1 8 0 -

6 - inline ull compensationc 1 7 5 -

L

9

0 _ I

1 7 0 -

630 pwhm da l rw dual dispersonE -15 5 I -800 -600 -400 -200 0 200 400 600 800 0 1 2 3 4 5

\

Total Accumulated Dispersion (pslnm) Span Count

Fig. 5 Maximum power Pmax= N Pd vs. totalresidual dispersion or experiment and simulation

Fig. 6 Evolution of the required OSNR with spancount for center channel in 9-channel experiment

product of launch power per channel and maxim um span count P,=Pd.N is constant. Th e validity of such amaximum power law has previously been demonstrated for self-phase-modulation[8]. Here, a P of 10.5 dBm isfound for a total accumulated dispersion of630 p s h m .

In our experimen ts with parallel polarization at the loop input we observe a deviation from the ma ximum pow erlaw when transmitting more than six spans (input parallel in Fig.4). This effect is attributed to the lossof thecorrelation between the polariza tions of the two channels. By scram bling the relative polarization of the channe ls at

the loop input the influence of the correlation loss on the mea surements is reduced since the worst case of relativeinput polarizations is dom inating (input scrambled in Fig.4). Simulation results for the two chann el transmission are also depicted in Fig.4. Consistent to the experiments a

maximum power law is found in the calculations. The value of P is different by1. 8 dB with respect to theexperiments which we attribute to the receiver modeling in the simulations.

The influenceof the residual dispersion on the maximum power P,,, is experimentally and numericallyanalyzed inFig. 5. Qualitative behavior agree well for simulation and experiment, whereas a quantitativediscrepancy is again due to receiver m odeling. The optimum dispersion v alue is slightly shifted from zerodispersion. From simulations neglecting SPM we find that this effect is not due to SPM . Rather,XPM generates thedominating phase distortions not exac tly at the fiber input but within one walk-off length in the fiber.

We also checked the validity of our results for a higher numberof WDM channels by launching ninewavelength channels with parallel polarization and100 GHz spacing into the loop. Whereas the center channel(reference) is modulated independently b y 4 s ow n transmitter, the PRBS data for the outer eight channels aregenerated by a smgle modulator and decorrelated by5 km of standard fiber.

Experimental results for the center channel are shown in Fig.4, using the same residual dispersion as for the2-channel results. The maximum power law is confirmed with aP which is reduced by5 .5 dB. The evolutionofthe required OSNR for a bit error rate of lo- is similar to the 2-channel case (Fig.6 ) . Again, appropriate inlinecompensation suppresses the build-up of theXP M penalty.

3. Summary

We have presented a detailed analysis of the impactof the dispersion map on the limits imposed by X PM on non-zero dispersion shifted fiber. From loop experim ents as well as simulations we find a maximu.m power law for theworst case w hich is given by100% nline dispersioncompensation.The systems performance, i.e. the maximumreach a nd the tolerance to residual dispersion of the w hole link, is substantially improved when using a slight under-compensation in each span. This finding also meets the requirements for optimum compensation schemes reducingself-phase modulation[8]. The qualitative behavior has been shown to be independent of the channel count.

4. References

.

[ l][2][3][4][5][6][7][8]

M. Eiselt,M. htaif, L.D. G arrett, OFC99, Tech. Dig., paper ThC-5, San Diego, 19 99 ,H.J. Thiele, R.I. Killey, P. Bayvel, ECOC98, Cod.Proc., Vol. 1, pp. 593-594, Madrid, 199 8[4]R. Hui, K.R. emarest, C.T. Allen,J. Lightw. Tech., vol. 17, no. 6, pp. 1018-102 6, 1999A. Cartaxo,J. Lightw. Tech., vol. 17, no. 2, pp. 178-190, 1999M. Varani, G. B elotti,A. Bononi,C. Francia, LEOS98, Conf.Proc., aper WBB4, Monterey, 1998A.J. Lucero,S . Ten, V.L. da Silva, OFC99, Tech. Dig., paper ThC-2, San Diego, 1999G..Bellotti et al. OFC2000, Postdeadline Paper PD32-1 ,2000J.-P. Elbers, A. Fahe rt, C. Scheerer, C. Glige ner,G. Fischer, EE E J. Sel. Top. Q uant. Electron,vo1.6,no.2, pp. :276-281, March 2000

I