Inventario de la red ciclista en Puebla

A key factor determining the amount o f XPM generated in a fibre link is the w alk-off which was shown in Fig. 2.15 as a function o f AT for different fibre types. In this section, XPM distortion is compared for the limit o f low and high walk-off. In the experiments, DSF and SSMF were used. For a given fibre dispersion, the channel spacing was additionally varied. Initially, the DSF was followed by dispersive fibre to carry out PM-IM conversion. This is a particularly suitable configuration to study fundamental properties o f XPM as the nonlinear section (DSF) is clearly separated from the linear phase to amplitude conversion (SSMF). Figure 4.6 shows the m easured time-averaged distorted probe waveform after transmission obtained by averaging the traces on the sampling oscilloscope.

SSMF versus DSF; The distortion measured for DSF fibre with low walk-off was compared to the case when SSMF fibre was used as the test-fibre. For AT=lnm, XPM induced distortion is reduced from mx=OAA for DSF to 0.11 for SSMF due to the increased w alk-off o f both channels over the first fibre from < 0 . 1 bit in DSF to «2 bits in SSMF. The sharp peak structure for DSF indicates localised build-up o f nonlinear chirp converted into IM by the following converting fibre. Therefore, the higher dispersion o f SSMF within L<Leff is advantageous as it reduces XPM distortion.

Chapter 4; XPM pump-probe experiments 79 V ex p er im en t sim u lation D

2

I

low w alk-off for /.</.,

D>0 and time time chirp; DSF+SSMF: _{SSMF only:} time PM-IM conversion:

Fig. 4.6 Probe distortion in pump-probe configuration. A 2=lnm , 13dBm/ehannel, (a) XPM after 45km DSF with 22km SSMF converter, Do,„v=370ps/nm, 2.5Gbit/s 1010...-modulation in pump, solid line: experiment, dashed line: split-step Fourier simulation, (b) (45+22)km SSMF, (c) qualitative explanation o f distortion for low and high channel w alk-off

Bit pattern: Next, the length of the 1010-... sequence o f the pump was varied by adding ‘1’- and ‘0’ to form longer clusters causing the temporal spacing between pulse transitions to increase from lOOps to 800ps. For the initial 1010...-sequence, the walk off over the DSF fibre was less than 0.1 bits. As a result, the peaks due to XPM distortion in the probe are clearly isolated similar to Fig. 4.6(a). From Fig. 4.7(a) to (d) the magnitude o f the XPM modulation index, expressed by rtix, remains constant, however, the standard deviation cxpm of the probe waveform increases for shorter pulse periods since more transitions occur within a given time interval. Therefore, cfxpm is more appropriate to quantify the total distortion o f the probe. In the limit o f low walk-off where neighbouring transitions are clearly isolated rrix remains constant and is only determined by the magnitude o f individual, separate transitions.

Chapter 4: XPM pump-probe experiments 80

As a result, rrix is expected to be independent on the modulation pattern under these circumstances and 1010.. .-modulation can be used instead o f a PRBS pattern.

0.9 0.9 0.9 0.8 0.8 time — ► 500ps

Fig. 4.7 Dependence o f XPM distortion on m odulation pattern in pum p at 2.5Gbit/s, 13dBm/channel, A /l=lnm , DSF+SSM F link, (a) 1010- (b) 110011- (c) 111 1000- (d) 8x1, 4x0- sequence

C hannel spacing: To investigate the influence o f channel spacing on XPM, rrix was measured as a function o f AT and this variation is shown in Fig. 4.8. The channel w alk-off increases with AT, filtering the pump channel as described by the transfer function H(co), and thus reducing the phase modulation o f the probe by the increased averaging over the waveform. XPM distortion o f m x> 0 .\ was detected for DSF over an interval o f more than ±10nm, whilst m^>0.05 was confined to AT<1.2nm for SSMF. The alternative use o f DCF as the converting fibre following DSF generated the same magnitude o f XPM distortion because o f the identical absolute dispersion. These measurements extend the results o f Fig. 4.6 and indicate that AT can be reduced in SSMF due to less XPM. The close channel spacing allows to increase the W DM transmission capacity. An interesting feature o f the DSF measurement was a local maximum, at approximately 6nm from AT=0 yielding an asymmetric plot with respect to AT=Onm. The increase o f rrix away from AT»0 is a result o f the w alk-off in DSF reaching a minimum when the zero dispersion wavelength To is situated exactly between pump and probe channel. This was also in good agreement with the simulations o f this experiment shown in Fig. 4.8. To avoid XPM even for wide channel spacing, careful consideration o f To was necessary during the system design. The graph in figure 4.8 became symmetrical either

Chapter 4; XPM pump-probe experiments 81

by placing the channels far away from Ào or by placing the probe channel directly at /lo~1555nm. Fig. 4.8 CO Û

0.2

0.6

0.15

0.4

0.2

0.05

-10

5

0

5

XPM distortion o f the probe as a function o f AÀ, (O ): SSM F, ( • ) : DSF, (— ): D SF simulation, 10 1 0 ...-m odulation, 2.5Gbit/s, 13dBm/channel, parallel polarisation