FIBER OPTICS

Prof. R.K. Shevgaonkar

Department of Electrical Engineering

Indian Institute of Technology, Bombay

Lecture: 33

Non-Linear Schrödinger Equation

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 1
 The two wave equations of pulse evolution of light in an optical fiber which were
introduced in the preceding discussion are:

{ ( ) ̃ } (33.1)

̃
(̃ )̃ (33.2)

The terms in the two equations retain their meanings. Since the frequency band of
optical signal is narrow, the quantity ̃ is very close to and so:

(̃ ) (̃ ) (33.3)

Substituting this approximation in equation (33.2) we obtain:

̃
(̃ )̃ (33.4)

Since we are concerned with a finite band of frequencies, the approximation of

equation (33.3) is not true everywhere and there is a small difference between the two.
Applying a Taylor series expansion around we have:
( )
( ) ( ) ( ) ( ) (33.5)

If we define:
( )
( ) ( ) (33.6)

Then equation (33.5) can be re-written as:

( )
( ) ( ) (33.7)

The dielectric constant of the optical fiber, which has both linear and non-linear
terms, is given by the square of the refractive index ‘n’ added with a negligibly small value
∆n due to the presence of non-linearity. The following approximation thus holds:

( ) (33.8)

As discussed earlier, the quantity ∆n can be expressed as a sum of the non-linear

and the loss terms as shown below:

|⃗ | (33.9)

The quantity ̃, which is a function of ω, can be expressed as:

̃( ) ( ) ( ) (33.10)

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 2
 The term ( ) can be obtained from the following expression containing the field
distribution function F.
( ) ∬ ( )| |
( ) (33.11)
( ) ∬| |

Here is the differential area of cross-section. Let us now write the expression for
the inverse Fourier transform of the envelope function of the signal. This expression is
given as:
( ) ∫ ̃( ) ( )
(33.12)

Substituting this expression in the equation (33.4) and solving, we obtain an

equation in the time domain given as:

| | (33.13)

The quantity is known as the non-linearity parameter and is related to the non-
linearity coefficient and the confinement of the light. It can be expressed as:

(33.14)

the term Aeff is the effective area of cross-section of the optical fiber within which the
light is confined and mathematically, this area can be expressed as:
(∬| | )
(33.15)
∬| |

The quantity F is the transverse field distribution inside the optical fiber and knowing
F enables us to determine the effective area A eff within which the light is confined in an
optical fiber. This, in turn, determines the non-linearity parameter from equation (33.14).
Depending upon the core size of the optical fiber, the field distribution etc. the typical
values of Aeff range between 1-100µm2. Correspondingly, the typical values for range
between 1-100 W -1/Km.

The equation (33.13) gives the behaviour of evolution of the optical signal envelope
with time as the signal propagates along the optical fiber. Let us know have a physical
insight into the equation.

If the optical fiber is loss-less ( ), linear ( ) and the quantity is negligibly

small, the resultant equation is given as:

(33.16)

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 3
 The quantity ⁄ is the group velocity ‘vg’ of the optical pulse inside the optical
fiber and is correctly given by equation (33.16) because the term is, in fact, the inverse
of the group velocity of the optical pulse inside the optical fiber (as can be derived from
equation (33.6)). Thus for a linear, loss-less optical fiber the optical pulse evolution
equation reduces to an equation expressing the group velocity of the pulse inside the
optical fiber.

Let us now assume that, we move along with the optical pulse with the group
velocity. In this case, the term ⁄ goes to zero, because there is no relative motion
between the observer and the pulse. This observation caters the need of defining a new
time frame with respect to a moving observer, moving at the group velocity. This new time
T can be expressed as:
(33.17)

Thus,

( ) (33.18)

Substituting this in equation (33.13), we obtain:

| | (33.19)
Equation (33.19) represents the Non-Linear Schrödinger equation (NLS). Thus the
NLS is the approximated differential equation governing the behaviour of the pulse
evolution of an optical pulse inside an optical fiber in presence of non-linear effects. Before
going into the solution of the NLS, let us first have a comprehensive physical
understanding of the NLS as we did in the preceding case.

Let us assume that, is negligibly small and the fiber is linear i.e. . The NLS,
hence, reduces to:

(33.20)
The above equation is a simple linear differential equation whose solution is:

( ) ( ) (33.21)

being the attenuation constant, the above equation suggests that the optical pulse
launched into the optical fiber would undergo an exponential decay w.r.t distance with an
attenuation constant . This situation is similar to any general lossy optical medium in
which the launched light decays exponentially. So, the term represents the loss in an
optical fiber. The terms on the RHS of equation (33.19) is the result of the presence of
non-linearity in the optical fiber material and the quantity is (as already discussed) the
non-linear parameter. The second term on the LHS of equation (33.19) is proportional to
which is, in fact, the rate of change of group velocity as a function frequency and
signifies that different frequencies travel with different group velocities. This phenomenon
of different wavelengths travelling with different velocities is known as dispersion. So, the

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 4
second term in the LHS of equation (33.19) basically represents dispersion in the optical
fiber.

In order to solve the NLS numerically, we define two operators as shown below:

̂ (33.22)

̂ | | (33.23)

One may note the fact that due to the presence of the differential operator, the
dispersion operator eases the solution of the equation in the frequency domain because a
differentiation in the time domain transforms to a frequency multiplication in the Fourier
transform domain. Similarly, the non-linear operator eases the solution of the NLS in the
time domain because of the presence of the envelope function A which is a time domain
function. These two notions suggest a very important fact that the NLS can be solved
relatively easily if it is solved simultaneously in the frequency as well as in time domains.
However, since both the non-linear and dispersion effects are inherently weak effects,
during the solution of the NLS for one the other effects may be assumed to be absent.
Considering one effect at a time also enables us to have a better physical insight to the
pulse evolution behaviour due to the individual effects.

Under the above assumptions, the pulse evolution is studied using the Split-Fourier
Step method. In this method the medium under study is subdivided into smaller sections
each having equal length ‘∆z’. The length chosen is such that the dispersion and the non-
linear effects remain weak effects and the consideration of one effect at a time is
adequate. Each subsection is then divided into two equal halves and the operator ̂ is
applied in the frequency domain until the spectrum reaches the end of the half. At this
point the time domain expression is obtained and the ̂ operator is applied to obtain the
pulse envelope at that point. Then for the next half, again the Fourier transform is taken
and the ̂ operator is applied until the spectrum reaches the end of the first subsection
where the ̂ is applied to the time domain expression. This trend is repeated until the
entire medium is spanned and the pulse evolution over the entire optical fiber can, thus, be
obatined. The figure below may help the reader to understand better:

Figure 33.1: Split Fourier Steps

Although analytical solutions of the NLS exist for certain specific pulse shapes,
complex pulse shapes can be solved numerically using the above method. Let us first
assume a Gaussian pulse with a pulse power of P and a standard deviation of T0. That is:
Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 5
 ( ) (33.24)

The characteristic dispersion length ‘LD’ over which the dispersive effects become
significant for the Gaussian pulse can be expressed as:

(33.25)
| |

The characteristic non-linearity length ‘LNL’ over which the non-linear effects
become significant for the Gaussian pulse can be expressed as:
(33.26)

Once the characteristic lengths are defined, four possible cases of the solution of
the NLS arise which are based on the actual physical length if the optical fiber under
consideration. If we consider a section of an optical fiber with a length L (and for simplicity
of comprehension let ), then the four possible cases are stated below:

Case 1:
In this case the chosen length of the optical fiber behaves just like a simple medium
to transport light from the input to the output without any significant modification occurring
in the pulse characteristics.

Case 2:
This scenario occurs in case of a low power, very narrow optical pulse. As the pulse
power is low, the non-linear effects are negligible and the resultant modification that occurs
to the pulse is the pulse broadening due to dispersion which has been already discussed
under linear optics. This situation is also known as the group velocity limited (GVD) regime
and most often this is the scenario that takes place in optical communication.

Case 3:
This situation occurs in case of a sufficiently broad pulse with high power so that
non-linear effects are significant. In this case, a phenomenon known as the self-phase-
modulation (SPM) is observed due to variation of the refractive index of the material by the
pulse shape itself. SPM shall be discussed in detail later. This regime is called as the Non-
Linearity limited regime.

Case 4:
This case arises when the pulse width is very narrow but there is high pulse power.
In this scenario, both dispersion as well as non-linear effects co-exists and we obtain a
very special kind of pulse propagation known as solitonic propagation.

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 6