QAM Transmitter

1 OBJECTIVE
Investigate the method for measuring the BER accurately and the distortions present in coherent
modulators.

2 PRE-LAB
The goal of optical communication systems is to transmit data from one point to another. Generally, this
data takes the form of a binary signal. The binary data needs to be used to modulate an optical signal
and then correctly demodulated at the receiver into a readable binary signal. Some simple schemes
have already been described, like the pervasive on off keying, where the optical power is used to
represent a ‘1’ or ‘0’. However, power is only one aspect of the electric field and the phase can also be
modulated. Describing the electric field with the complex envelope makes this clear:

𝑬𝒕 (𝑡) = 𝑅𝑒[𝐸𝑥 𝑒 𝑖𝜙𝑥 −𝑖𝜔𝑡 𝑥̂ + 𝐸𝑦 𝑒 𝑖𝜙𝑦 −𝑖𝜔𝑡 𝑦̂]. (1)

From this equation, it is evident that for each polarization the amplitude can be changed and also the
phase, 𝜙𝑥 . Amplitude modulation is a more general case of on-off keying (OOK) where the amplitude
may take on more than two different levels. Modifying the phase, called phase shift keying (PSK), does
not have an effect on the power of the output, but the phase of the optical carrier. Then there is the
logical next step to modulate the amplitude and the phase at the same time called Quadrature
Amplitude Modulation (QAM).

a) b)

1
 Figure 1: An example of two different yet similar modulation formats. Figure a) represents an on off keying non return to zero
format, while Figure b) shows a return to zero format.

2.1 BIT RATE, SYMBOL RATE AND BIT ERROR RATE

At this point the reader should be fairly confident to describe what the bit rate is. In any case, the bit
rate is defined as the number of bits per second transmitted. The bit period, 𝑇𝐵 , and in the special case
of NRZ also the pulse width, would be given by:

𝑇𝐵 = 1⁄𝐵 (2)

In advanced modulation schemes instead of only having two states (on or off) there might be 4 or even
64 different states. The term bit is then replaced with symbol and bit rate replaced with symbol rate.

𝑆 = 𝐵⁄𝑙𝑜𝑔 𝑀 (3)
2

Here S is the symbol rate and M is the number of states that a symbol can take. For example, a system
where the amplitude can be 0, 0.5, 1, 1.5 would have M = 4 and result in a symbol rate half that of a
regular OOK format.

The bit error rate (BER) is an extremely useful metric for describing the quality of communication
system. Simply put it is the number of misinterpreted bits per bit sequence transmitted.
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑏𝑖𝑡 𝑒𝑟𝑟𝑜𝑟𝑠
𝐵𝐸𝑅 = (4)
𝑙𝑒𝑛𝑔𝑡ℎ 𝑜𝑓 𝑏𝑖𝑡 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒

The goal of communication systems is to keep this value low, while increasing the bit rate and at the
same time keeping pecuniary costs low. This means that an accurate calculation of the BER depends on
the length of the bit sequence. Suppose the true BER of a system is quite low, one error in a million bits
or a BER of 10-6, a simulation or laboratory test with only 1000 bits would only have a 0.1% chance of
encountering a single bit error. Assuming that the experiment did result in an error, the BER would be
incorrectly calculated as 10-3, which is of course far off from the true BER.

The BER can also be estimated by inspecting the eye diagram from a NRS/RZ type format and for more
advanced formats the constellation diagram can be used. In both of these cases though the longer the
bit sequence the more accurate the results.

Build the following layout in OptiSystem to investigate the sequence length’s effect on the BER.

2
 Figure 2: Layout for calculating BER of NRZ.

Once the components have been placed change the Power of the Optical Transmitter to -10 dBm and
the Symbol rate, found in the layout parameters, to 10e+009 symbols/s

Questions:

2.1.1 Calculate the average BER and standard deviation from the Min. BER result given by the
BER Analyzer for 25 identical simulations. The random noise and exact bit sequence will
be the only changes.

2.1.2 Change the sequence length to 1024 and calculate the average BER and standard
deviation from the Min. BER result given by the BER Analyzer for 25 identical
simulations.

2.1.3 Change the sequence length to 4096 and calculate the average BER and standard
deviation from the Min. BER result given by the BER Analyzer for 25 identical
simulations. You may need to uncheck the BER Analyzer option “Eye must be open”,
since the eye becomes crowded from the numerous bits.

2.1.4 What is the effect of the sequence length on the calculated BER?

2.2 MODULATING LIGHT

Understanding the modulation scheme of OOK is quite simple, the receiver simply detects the power
levels of signal and decodes it as a one or a zero. The important question is how to translate the
electrical signal to an optical signal. In some simple systems, the pump current to the laser is modulated
to change the intensity. This scheme is called direct modulation and normally results in pulses that are
chirped from induced change in refractive index due to the change in carrier density. As chirped pulses
are strongly affected by dispersion in optical fibers, directly modulated lasers are rarely used. A better
method is to keep the laser at a steady current and modulate the signal externally with an alternate
device.

A Mach Zehnder Modulator (MZM) can be used to modulate both the amplitude and phase of an optical
signal. Without getting too detailed, the MZM splits the optical signal into two branches and induces a
phase shift on each branch independently. At the output, the two signals are recombined and can result
in a change in amplitude or phase. The response is given by:

3
 𝜋𝑉 𝜋𝑉2
𝐸𝑖𝑛 𝑗( 1) 𝑗( )
𝐸𝑜𝑢𝑡 = 2
[𝑒 𝑉𝜋 +𝑒 𝑉𝜋 ], (5)

where 𝑉1is the applied voltage at the top branch and 𝑉2 is the applied voltage at the bottom branch. 𝑉𝜋 is
the required voltage to induce a PI phase shift to one of the arms. By varying the voltage on each arm,
the amplitude can be modulated via constructive or destructive interference. By applying the same
voltage to the top and bottom arms the phase can be changed as well.

Questions:

2.2.1 What does the MZM response simplify to in the case where 𝑉2 = −𝑉1?

2.2.2 In the previous expression, consider the case where 𝑉1 = 𝑉(𝑡) + 𝑉𝑏𝑖𝑎𝑠 , what value of
𝑉𝑏𝑖𝑎𝑠 should be used such that the response is close to linear for small 𝑉(𝑡).
Hint: sin 𝑥 ≈ 𝑥 for 𝑥 → 0

2.3 QUADRATURE AMPLITUDE MODULATION

A special type of modulation which can easily be expanded to carrying many bits per symbol is
quadrature amplitude modulation (QAM). At its core, it is the combination of modulating the phase as
well as the amplitude and is expressed by the complex signal:

𝑬(𝑡) = 𝑅𝑒[(𝐼(𝑡) + 𝑖𝑄(𝑡))𝑒 𝑖𝜔𝑡 ]. (6)

One can take the view point that 𝑄(𝑡) modulates the imaginary component of the complex envelope
while 𝐼(𝑡) modulates the real part of the complex envelope. Removing the complex notation it
becomes:

𝑬(𝑡) = 𝐼(𝑡) 𝑐𝑜𝑠 𝜔𝑡 − 𝑄(𝑡) 𝑠𝑖𝑛 𝜔𝑡. (7)

The 𝑄(𝑡) signal is out of phase by 90 degrees and gets the name quadrature phase carrier, while the
other is the in-phase carrier. To visualize these different symbols, they are plotted on a complex plane or
constellation. For example, a 4 QAM signal has two bits per symbol and 𝐼(𝑡) and 𝑄(𝑡) can both take
values of 1 and –1, representing a bit value 1 and 0. By removing the time domain from the equation, a
constellation diagram can be used to view which symbols are present. For example, take Figure 1 below
which is a representation of a 4 QAM signal. The binary sequence is cut into 2 bit length symbols where
the first bit is translated for the 𝐼(𝑡) component and the second is translated for the 𝑄(𝑡) component.
To create the final signal, all that is needed is a way to modulate two sinusoids that differ in phase. To
accomplish this, a CW laser can be split by a 3 dB cross coupler, which by its nature induces a π/2 phase
shift.

4
 01 11

00 10

Figure 3: Constellation diagram of 4 QAM signal. Binary signals cut into 2 bit long symbols.

The following diagram demonstrates how a coherent transmitter could be designed. The binary
sequence is split into and subsequently converted to the electrical signals, 𝐼(𝑡) and 𝑄(𝑡). The electrical
signals modulate their respective carriers, which differ by a π/2 phase shift due to the cross coupler.

𝐼(𝑡)

𝑄(𝑡)

Figure 4: QAM transmitter design

2.3.1 The MZMs are held at the null bias point or a total bias voltage equal to 𝑉𝜋 . This allows
the time dependent portion of the signal to sweep through the maximum range of
values before the periodic nature of the response duplicates values. What should the
gain be if the M-ary Pulse Generator is outputting pulses of -1 and 1 Volts? What about
a signal that ranges between -5V, -3V, -1V, 1V, 3V, 5V? Assume a 𝑉𝜋 of 3 V.

2.3.2 In the constellation diagrams below convert the symbols in to bit sequences by the
following protocol. First find the number of bits per symbol. The first half of the bits will

5
 represent the 𝐼(𝑡) amplitude and the last half will represent the 𝑄(𝑡) amplitude. Then,
number the possible amplitudes starting from the smallest to the largest for each axis.
Finally, locate the highlighted symbol and represent it as a binary sequence.

a) b) c) d)

Example: In a) there are 16 possible states, which can be represented by 4 bits.

The first two bits will represent the possible amplitudes for the in phase
portion while the last two will represent the quadrature phase.

Bit Sequence I(t) Q(t)

0000 ‘00’–> –3 ‘00’–> –3
0001 ‘00’–> –3 ‘01’–> –1
0010 ‘00’–> –3 ‘10’–> 1
0011 ‘00’–> –3 ‘11’–> 3
0100 ‘01’–> –1 ‘00’–> –3
0101 ‘01’–> –1 ‘01’–> –1
0110 ‘01’–> –1 ‘10’–> 1
0111 ‘01’–> –1 ‘11’–> 3
1000 ‘10’–> 1 ‘00’–> –3
1001 ‘10’–> 1 ‘01’–> –1
1010 ‘10’–> 1 ‘10’–> 1
1011 ‘10’–> 1 ‘11’–> 3
1100 ‘11’–> 3 ‘00’–> –3
1101 ‘11’–> 3 ‘01’–> –1
1110 ‘11’–> 3 ‘10’–> 1
1111 ‘11’–> 3 ‘11’–> 3

In this labelling scheme the circled symbol results in the bit sequence ‘0111’.

6
3 DISTORTION IN QAM TRANSMITTERS
Build a QAM transmitter from the layout shown in Figure 3 and after creating a subsystem out of the
components attach a bit sequence generator and a pre-built Optical Coherent QAM Receiver as in the
diagram below:

Figure 5: Layout of QAM transmitter.

 Pseudo-Random Bit Sequence Generator Transmitters Library/Bit Sequence Generators

 QAM Sequence Generator Transmitters Library/Coders
 M-ary Pulse Generator Transmitters Library/Pulse Generators/Electrical
 Fork Tools Library
 CW Laser Transmitters Library/Optical Sources/
 Electrical Gain Signal Processing/Arithmetic/Electrical
 Electrical Bias Signal Processing/Arithmetic/Electrical
 LiNb Mach-Zehnder Modulator Transmitters Library/Modulators/Optical
 Phase Shift Passives Library/Optical
 X Coupler Passives Library/Optical/Couplers
 Optical Coherent QAM Receiver Receivers Library/Optical Receivers
 Electrical Constellation Visualizer Visualizer Library/Electrical

The DC bias point for the two MZMs will be the null point or 𝑉𝜋 and in this case that corresponds to a
1.5V bias at the top port and a -1.5V on the bottom port. In the LiNb MZMs make sure to change the
Extinction Ratio to 25 dB, both switching voltages to 3V and the insertion loss to 1 dB. The X Couplers
should also include a 1 dB insertion loss. The CW Lasers in this project are also slightly different as in
coherent transmission links you need better phase characteristics of the transmitter. For the two CW
Lasers change the Linewidth to 0.1 MHz and set them to operate at 193.1 THz. The optical coherent
receiver can be set to accept an external local oscillator (laser), a necessary additional component in
coherent receivers that will be explained in a later laboratory. At this stage the assumption that both

7
lasers have the exact same phase characteristics will be made to simplify the design. In the Component
Properties of the two CW Lasers change the Random seed index to 0 for both.

3.1 EFFECT OF GAIN

The constructed QAM Optical Transmitter can be modified easily to obtain both 4 QAM and 64 QAM,
first set the transmitter to function as a 4 QAM transmitter by setting the Bits per symbol appropriately.

Questions:

3.1.1 Assuming the default bit rate of 10 Gbps, what should the Layout Parameter Symbol
rate be set to for 4 QAM and 64 QAM?

3.1.2 What is the maximum gain that should be used for the case of 4 QAM and 64 QAM?

3.1.3 Simulate the effect of the gain on the 4 QAM constellation at the output starting from
the maximum gain value and ending at 0.1. Compare the first and final constellations
and describe what the effect of the gain on the constellation. There will be some
random noise generated by the photodetectors in the optical receiver.

3.1.4 The off center characteristic of the constellation is due to an additional term in the
transfer function of the MZMs, that was not explained in the pre-lab but has an
important effect nevertheless. Find the transfer function in the technical background
and explain which parameter is responsible for this behavior.

3.1.5 Simulate the effect of the gain on the 64 QAM constellation at the output starting from
the maximum gain value and ending at 0.1. Compare the first and final constellations
and describe what the effect of the gain on the constellation.

4 REPORT
In your lab report include the following:

 Brief overview of the background and theory.

 Answers to all pre lab questions, clearly showing your work.
 Brief description of the simulation method and setup, including screenshots.
 Final results including figures and discussion.

5 REFERENCES
[1] Agrawal, G. P. Fiber-optic Communication Systems. New York: Wiley, 1997. Print