Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Chapter 4

Chapter 4 GVD and SPM-induced degradation effects on a Gaussian pulse

GVD AND SPM-INDUCED DEGRADATION EFFECTS ON A GAUSSIAN PULSE


This chapter contains mathematical analysis and simulation results of the effects of group velocity dispersion and self-phase modulation on a Gaussian pulse propagating through a single mode optical fiber. Numerical model based on split-step Fourier technique, as discussed in chapter 3, was used to obtain the results. This work presents more detailed understanding of GVD and SPM effects on the propagating pulse and provides an opportunity to compare the results obtained from the numerical model with some well-known results available in the literature.

4.1 Dispersion-Induced Pulse Broadening


The pulse propagating through optical fiber is affected both by GVD and SPM simultaneously. However each of these effects can be investigated alone by setting such values of pulse-width and the peak power so that the other effect can be ignored, as mentioned in (Agrawal, 2001) and also discussed briefly in section 4.3. This section presents an analysis of dispersion-induced pulse broadening by ignoring SPM effects. The effects of SPM will be considered in section 4.2. For these analyses, the approach given in (Agrawal, 2001; Agrawal, 2003) will be used. The GVD effect alone on pulse propagation can be investigated by setting = 0 in equation (3.95) that leads to
i
A i 1 2 A = A + 2 2 . z 2 2 T

(4.1)

A normalized amplitude U(z,T) can be introduced through the following expression 53

z A( z , T ) = P0 exp U ( z, T ) , 2

(4.2)

where P0 is the peak power of the pulse incident at the input end of the fiber. Using equation (4.2), the following expressions can be obtained
A z = P0 U exp z 2 2 z U , + P0 exp 2 z

(4.3)

and
2 2 A z U = P0 exp 2. T 2 2 T

(4.4)

Using equations (4.2)-(4.4) in (4.1) leads to


z i P0 U exp 2 2

i z U + P0 exp = 2 2 z

z P0 exp U + 2
2 z U P0 exp 2, 2 T

2
2
i
U 2 2U = . z 2 T 2

(4.5)

The linear partial differential equation (4.5) can be solved using the method of Fourier transform as given below. Denoting U ( z, ) as Fourier transform of U(z,T) such that
U ( z, T ) =

1 U ( z, ) exp ( iT ) d . 2

(4.6)

Using equation (4.6), the following terms are obtained,


1 U U exp ( iT ) d , = z 2 z 1 U = ( i) U exp ( iT ) d , T 2 1 2U = U exp ( iT ) 2 d . 2 2 T

(4.7)

(4.8)

54

Substitution of equations (4.7) and (4.8) into (4.5) leads to


i
1 1 U 2 exp ( iT ) d = 2 U exp ( iT ) d , 2 z 2 2

U = 2 2U . z 2

(4.9)

By rearranging the terms, the following expression is obtained


U i = 2 2 z. U 2

(4.10)

Integrating and simplifying both of the sides of (4.10) leads to

ln U

U ( z , ) U ( 0, )

i = 2 2 z , 2 0
(4.11)

i U ( z , ) = U (0, ) exp 2 2 z , 2

where U (0, ) is the Fourier transform of the incident field at z = 0 which can be calculated from U(0,T) as

U (0, ) =

U (0, T ) exp(i) dT .

(4.12)

Equation (4.11) shows that phase of each spectral component of the pulse changes due to GVD. The amount of phase change depends on both the frequency and the propagated distance in optical fiber. Consequently, the shape of the pulse is modified. Using (4.11) in (4.6), the general expression for the resultant field becomes as
U ( z, T ) = 1 2

U (0, ) exp 2
2

z iT ] d .

(4.13)

For a Gaussian input pulse, given as

T2 U (0, T ) = exp 2 , 2T0

(4.14)

55

where T0 represents the half-width at 1/e intensity point. In practice, full-width half maximum (FWHM) term is commonly used which can be related to T0 by using the intensity defined as
U (0, T ) = U *U = I ,
2

(4.15)

where I stands for intensity. Substituting (4.14) into (4.15) leads to

T2 I = I 0 exp 2 , T0

(4.16)

where I0 = 1. Half-width at half maximum can be obtained by setting T = THWHM and I = I0/2 in equation (4.16) so that

T2 I0 = I 0 exp HWHM 2 T02


or

THWHM = ( ln ( 2 ) ) 2 T0 .

(4.17)

Thus full-width at half maximum is


TFWHM 1.665T0 .

(4.18)

Using (4.14) into (4.12) leads to

U (0, ) =

T2 exp 2 exp(i)dT , 2T0


T2 exp 2 iT dT . 2T0
+

U (0, ) =

(4.19)

Using the following identity


+

2 exp ( ax + bx + c ) dx =

b 2 4ac exp , a 4a

(4.20)

equation (4.19) takes the form as

56

2T02 U (0, ) = 2 T0 exp . 2


Using (4.21) into (4.13) leads to

(4.21)

U ( z, T ) =

1 2

2T02 i 2 2 T0 exp exp 2 z iT d , 2 2


T 2 i exp 0 2 z 2 + iT d . 2 2
+

U ( z, T ) =

T0 2

(4.22)

Again using the identity given in (4.20) with a = (4.22) takes the form

T02 i 2 z , b = iT , c = 0 , equation 2

U ( z, T ) =

T0 2

2 ( iT ) 0 , exp T02 i 2 z T02 i 2 z 4 2 2


T0

U ( z, T ) =

T2 exp . 2 2 T0 i2 z 2(T0 i2 z )

(4.23)

This expression clearly shows that in the presence of GVD alone, the input pulse maintains its Gaussian shape during propagation, however it experiences broadening that increases with the propagated distance z. Using equation (4.23), the half-width of the pulse (Tz) at 1/e intensity point at any distance z along the length of the fiber can be related to half-width of the incident pulse (T0) at 1/e intensity point by the following procedure. Defining dispersion length LD (which governs the extent of broadening of the optical pulse) as
LD = T02 ,

(4.24)

57

and using equations (4.23) and (4.15), the expression for U ( z, T ) in terms of LD can
2

be obtained as

U ( z, T ) =
2

T2 1 exp 2 . 1/2 2 1 + ( z / LD ) 2 T0 1 + ( z / LD ) 1

(4.25)

Equation (4.25) can be written as


T2 1 I = I 0 exp 2 , 2 T0 1 + ( z / LD )

(4.26)

where
I0 = 1 1 + ( z / LD ) 2
1/2

If I =

I0 , then for this case denoting T as Tz, equation (4.26) takes the form e

T 1 1 , = exp z2 2 e T0 1 + ( z / LD )

z 2 Tz = T0 1 + . LD

1/2

(4.27)

Equation (4.27) shows that a propagating Gaussian pulse will be broadened by a factor of

2 at z = LD. Equation (4.23) can be represented in terms of LD as


U ( z, T ) = 1 T2 exp 2 . 1 iz / LD 2T0 (1 iz / LD ) (4.28)

Using equation (4.28), the dispersion-induced broadening of a Gaussian pulse at distances z = LD, 2LD, and 3LD is illustrated in Figure 4.1. It shows that the pulse experiences broadening due to GVD. The broadening goes on increasing as the propagated distance increases in the fiber. It is pointed out that in this section of the thesis, for the analysis of fiber dispersion alone, fiber loss is not considered and
58

therefore all the curves of Figure 4.1 carry the same amount of net energy. The pulse spreads on the expense of reduction in peak power.
1 z / LD = 0 0.9

0.8

z / LD = 1

0.7

0.6 |U(z,T)|2

z / LD = 2

0.5 z / LD = 3

0.4

0.3

0.2

0.1

0 -15

-10

-5

0 T/T0

10

15

Figure 4.1 Dispersion-induced broadening of a Gaussian pulse at different propagated distances

The comparison of equations (4.14) and (4.23), which represent the incident field and the resultant field respectively, shows that even when the incident pulse is un-chirped, the transmitted optical pulse becomes chirped due to fiber dispersion. To elaborate this fact, U(z,T) can be written in the form

U ( z , T ) = U ( z , T ) exp [i ( z, T )] ,
where

(4.29)

( z, T ) =

sgn ( 2 )( z / LD ) T 2 1 + tan 1 ( z / LD ) . 2 2 T0 2 1 + ( z / LD )

(4.30)

Equation (4.30) shows that the instantaneous frequency differs across the pulse from the central frequency 0 due to time dependence of the phase ( z , T ) .

59

The difference is given as

(T ) =

sgn ( 2 )( 2 z / LD ) T , = 2 T T0 2 1 + ( z / LD )

(4.31)

which shows that fiber dispersion imposes linear frequency chirp on the pulse. The equation (4.27) and results shown in Figure 4.1 clearly indicate that GVD broadens the width of the incident pulse.

4.2 SPM-Induced Spectral Broadening


The phenomenon of SPM was introduced in section 2.3. It is the dominant nonlinear effect that limits the performance of single channel light wave systems by giving rise to an intensity dependent phase shift which introduces chirp in the propagating signal and hence causes spectral broadening. In a long-haul communication link, the SPM effect accumulates due to inline amplifiers which are used periodically for loss compensation. The effect of SPM alone on pulse propagation can be investigated by setting 2 = 0 in pulse propagation equation (3.95) that takes the form as
i A i 2 + A + A A = 0. z 2

(4.32)

Using the normalized amplitude U(z,T) defined in equation (4.2), re-written below
z A( z , T ) = P0 exp U ( z, T ) , 2

(4.33)

and differentiating both sides of equation (4.33), the following expression is obtained

A( z , T ) z z U ( z, T ) , = P0 exp . U ( z, T ) + P0 exp . z z 2 2 2
A( z , T ) U z = i P0 exp U + . z z 2 2

(4.34)

Using equations (4.33) and (4.34) in (4.32) leads to

60

U i z z i P0 exp P0 exp U + + U + z 2 2 2 2 z P0 exp P0 exp ( z ) U 2 U = 0, 2

U i exp ( z ) 2 = U U, z LNL

(4.35)

where LNL is nonlinear length that depends on peak power P0 and the nonlinear parameter as
LNL = ( P0 ) .
1

(4.36)

Consider the following substitution


U = V exp(i NL ) ,

(4.37)

and hence U V = exp(i NL ) + iV exp(i NL ) NL . z z z Using equations (4.37) and (4.38) in (4.35) leads to V exp( z ) 2 exp(i NL ) + iV exp(i NL ) NL = i V V exp(i NL ). z z LNL (4.39) (4.38)

Equating the real and imaginary parts in equation (4.39), leads to the following two equations
V = 0, z

(4.40)

and

NL exp ( z ) 2 V . = LNL z

(4.41)

Equation (4.40) shows that V is independent of the propagated distance z and thus in the presence of SPM alone, the amplitude of the pulse remains unchanged as it propagates through fiber. Hence equation (4.37) can be integrated analytically between the limits 0 L to have the general solution as 61

U ( L, T ) = U (0, T ) exp [iNL ( L, T )] ,

(4.42)

where V = U ( 0, T ) represents the field amplitude at z = 0 and NL is nonlinear phase shift that can be determined as given in the following. Integrating equation (4.41) between the limits 0 L leads to

=
NL

V 2 exp( z ) , LNL 0
(4.43)

=
NL

V 2 1 exp( L) . LNL

As V 2 = U ( 0, T ) , therefore equation (4.43) becomes


2

U ( 0, T ) 1 exp( L) NL = , LNL
2

NL ( L, T ) = U (0, T ) ( Leff / LNL ) ,


where Leff is the effective length defined as
Leff = 1 exp( L)

(4.44)

(4.45)

In the absence of fiber loss, Leff becomes equal to L. However in the presence of fiber loss, the SPM effect goes on decreasing across the fiber length and thus Leff < L. Equations (4.42) and (4.44) show that SPM gives rise to an intensity dependant phase shift. The nonlinear phase shift NL increases with increase in the propagated distance. Since the maximum phase shift max occurs at the pulse centre that is located at T = 0, hence equation (4.44) leads to

max = U (0, 0) ( Leff / LNL ) .

(4.46)

Considering the normalized amplitude U i.e. |U(0,0)| = 1 and also using (4.36), equation (4.46) becomes 62

max =

Leff LNL

= P0 Leff , Leff

LNL =

max

(4.47)

The physical meanings of nonlinear length LNL is clear from equation (4.47) which shows that LNL is the effective propagation distance at which max = 1. Equation (4.44) shows the time dependence of NL that causes SPM-induced spectral broadening. Due to the temporally varying phase, the instantaneous optical frequency differs across the pulse from its central value 0 . The difference (T ) can be determined using equation (4.44) as given below

(T ) =

L NL 2 U (0, T ) . = eff T LNL T

(4.48)

The time dependence of is known as frequency chirping (the instantaneous frequency is time dependent in a chirped signal). If the frequency increases with time the chirp is called positive chirp or up-chirp (shown in Figure 4.2) and vice versa.

Electric field (arb. units)

-1

0 Time (arb. units)

Figure 4.2 Illustration of frequency up-chirp

63

The SPM-induced chirp increases in magnitude as the pulse propagates in the fiber. It implies that new frequency components are continuously generated during pulse propagation in the fiber which broaden the spectrum of the propagated pulse as compared to the spectrum-width of the incident pulse at z = 0. The extent of SPMinduced frequency chirp for a Gaussian pulse can be determined using equations (4.14) and (4.48), given by

(T ) =

T2 2 Leff T exp 2 . T0 LNL T0 T0

(4.49)

Nonlinear phase shift at Leff = LNL determined from equation (4.44) is given below

NL ( L, T ) = U (0, T ) .

(4.50)

Using equations (4.50) and (4.49), the variation of NL and the induced frequency chirp across the pulse at Leff = LNL for a Gaussian pulse are shown in Figures 4.3 and 4.4 respectively.
1

0.8

Phase (

NL

0.6

0.4

0.2

0 -3

-2

-1

0 T/T

Figure 4.3 Temporal variation of SPM-induced phase shift

64

0.8

0.6

0.4 Frequency chirp ( To)

0.2

-0.2

-0.4

-0.6

-0.8

-1 -3

-2

-1

0 T/To

Figure 4.4 Temporal variation of SPM-induced frequency chirp

There are some important points to be noted in Figures 4.3 and 4.4 as (i) Phase shift is directly proportional to the intensity of the pulse according to equation (4.50). Therefore, its temporal variation is identical to that of the pulse intensity, (ii) The temporal variation of SPM-induced chirp shows that is negative near the leading edge and becomes positive near the trailing edge of the pulse, (iii) The chirp induced by SPM is linear over a large central region of the Gaussian pulse.

4.3 Interaction of GVD and SPM


In sections 4.1 and 4.2, the effects of GVD and SPM were considered separately, i.e. neglecting one effect while investigating the other. In practice, GVD and SPM effects occur simultaneously during pulse propagation in optical fiber. The interplay of these effects produces interesting and different results than the results obtained in the presence of either of the phenomenon alone. In this section, the effects of interaction of GVD and SPM are discussed. 65

The following expression introduces a time scale normalized to the half-width T0 (at 1/e intensity point) of the initial pulse as = T t z / vg = . T0 T0 (4.51)

Using (4.33) and (4.51), the pulse propagation equation (3.95) can be represented as i U sgn ( 2 ) 2U exp ( z ) 2 = U U, z 2 LD 2 LNL (4.52)

where sgn(2) is +1 for normal and -1 for anomalous dispersion regime. Using the normalized distance defined as

=
and introducing a parameter N as
N2 =

z , LD

(4.53)

LD P0T0 2 = , LNL 2

(4.54)

equation (4.52) can be written in the form

U sgn ( 2 ) 2U 2 = N 2 exp ( z ) U U . 2 2

(4.55)

The parameter N governs the relative importance of the SPM and GVD effects on the optical pulse propagating through the fiber. Split-step Fourier method was used to implement equation (4.55) numerically. Equal role of GVD and SPM during pulse evolution can be considered by setting the parameters in equation (4.54) such that N ~ 1. The effect of interaction of GVD and SPM depends on the sign of 2 as clear from equation (4.55), therefore the pulse evolution behavior will be different for 2 > 0 (normal dispersion regime) and 2 < 0 (anomalous dispersion regime). The results of interaction of GVD and SPM for N =1 for both normal and anomalous dispersion regimes are discussed below.

66

4.3.1 Normal Dispersion Regime


Figure 4.5 shows evolution of the optical pulse in normal dispersion regime for N = 0 (means GVD alone) and N = 1 (means GVD and SPM both) for a lossless fiber ( = 0) for an initially un-chirped Gaussian pulse. Since SPM effect depends on the pulse power which goes on decreasing due to fiber loss, hence the results of interaction of GVD and SPM will be different in the presence and absence of fiber loss and are discussed in section 4.4.
D only D+ Input pulse 10

8 Intensity (mW)

2 5 0 4 3 2 z/LD 1 0 -10 -5 0 T/To

10

Figure 4.5 Comparison of pulse evolution with GVD alone, due to interaction of GVD and SPM in normal dispersion regime

It is clear from Figure 4.5 that the pulse broadens more for N = 1 (in the presence of GVD and SPM both) in comparison to N = 0 (i.e. GVD alone) for the same distance of propagation. It is due to the fact that SPM-induced chirp generates new frequency components, as discussed in section 4.2. These frequency components are red shifted near the leading edge and blue shifted near the trailing edge of the pulse. As in normal dispersion regime, red components travel faster than the blue components, therefore 67

the two chirps introduced by GVD and SPM support each other that lead to an enhanced rate of pulse broadening in comparison to that expected from GVD alone.

4.3.2 Anomalous Dispersion Regime


The pulse evolution behavior of the optical pulse propagating in anomalous dispersion regime in the presence and absence of nonlinearity for lossless fiber ( = 0) for an initially un-chirped Gaussian pulse is shown in Figure 4.6.
D+ 10 Input pulse D only

Intensity (mW)

2 15 0 4 3 2 z / LD 1 0 -15 -10 -5 0 T / To 10 5

Figure 4.6 Comparison of pulse evolution with GVD alone, SPM alone and their co-existence in anomalous dispersion regime

In this regime, the pulse broadens at a lower rate for the case of N =1 (both GVD and SPM) in comparison to N = 0 (i.e. GVD alone). It is due to the fact that SPM-induced chirp is positive while GVD-induced chirp is negative in anomalous dispersion regime, hence they compete each other resulting in less broadening than for the case of GVD alone. The results for lossy fiber will be provided in section 4.4. Modern long-haul fiber optic communication systems operate in the 1550 nm window of SMF where most fibers exhibit anomalous dispersion behavior. Therefore, the 68

remaining dissertation will focus on the anomalous dispersion regime of single mode fiber. Also the words fiber and SMF will be use synonymously.

4.4 Interplay of Loss, GVD and SPM


This section considers the interplay of GVD and SPM in the presence of loss in a fiber span of 80 km long with attenuation coefficient (), GVD (D) and nonlinear coefficient () as 0.22 dB/km, 16 ps/(nm.km) and 1.37 W-1/km, respectively. A Gaussian pulse having 8 dBm (~6.3 mW) peak power and full-width half maximum of 65 ps is propagated in the fiber. Figure 4.7 shows the degradation effects for thee cases: (i) due to GVD alone, (ii) due to interaction of GVD and SPM in the absence of fiber loss, and (iii) due to interaction of all these three effects. For case (iii), the net loss was fully compensated at the output side of the fiber.
7

6
Input pulse

Case (ii) D + Case (iii) + D + Case (i) D only

Power (mW)

0 -150

-100

-50

0 t (ps)

50

100

150

Figure 4.7 Broadening of a Gaussian pulse for different cases in SMF span of 80 km long

It is clear from Figure 4.7 that the increase in the pulse-width is the highest when only the dispersion effect is considered. Inclusion of SPM effect reduces the pulse spread. 69

The reason for this phenomenon has already been described in section 4.2 that the chirps introduced by GVD and SPM oppose each other resulting in minimum pulse degradation. The figure also shows that the inclusion of fiber loss causes the pulsewidth to increase than lossless case. It happens due to the fact that fiber loss attenuates the pulse power and therefore the effect of SPM keeps on decreasing resulting in dominant role of fiber dispersion which leads to more pulse spread. Figure 4.8 illustrates output pulses of Figure 4.7 normalized to their corresponding maximum values to show a clear cut comparison of pulse spread for each of the cases.
1 0.9 0.8 0.7 0.6 |U(z,T)|2 0.5 0.4 0.3 0.2 0.1 0 -150 D D + gamma data3 alpha++Gamma

Case (iii) + D + Case (ii) D + Case (i) D only Input pulse

-100

-50

0 t (ps)

50

100

150

Figure 4.8 Broadening of a Gaussian pulse shown in normalized form

These results confirm the well known fact that all these effects must be considered simultaneously in a real optical fiber link.

4.5 Summary
Mathematical and simulation analysis of group velocity dispersion and self-phase modulation were presented. The evolution of the pulse was discussed in the presence of GVD and SPM individually and collectively in the presence and absence of fiber 70

loss. The results presented in this chapter were obtained through a numerical model of pulse propagation and were found consistent with the results already given in the literature. These results have also shown the well known fact that it is necessary to consider these phenomena simultaneously during pulse propagation through optical fiber. On the basis of these results, chapter 5 shows the importance of optimization of dispersion-map and launch power to minimize the combined effects of fiber loss, GVD and SPM.

71

You might also like