Final Report of Telecommunication Engineering Training

Suzuki Laboratory, Mobile Communication Group

Tokyo Institute of Technology, Japan
Chapter 7

Lecture Note Developing on Simulation Programming

of 16QAM in the AWGN Channel

Author: Tri Budi Santoso

Supervisors: Prof. Hiroshi SUZUKI
Prof. Kazuhiko FUKAWA
Book Reference:
1. Digital Communication by Satellite
2. Digital Communications, John G. Proakis
Topic: 16QAM Transmission System

7.1. Session 1: Review of 16QAM

7.1.1. The Signaling of 16QAM
In M-ary PSK the signal amplitude is the same for all signals thus constraining the signal
points to a circular constellation. If this constraint is removed and the in-phase (I) and
quadrature (Q) signal components are allowed to vary independently we have a scheme called
quadrature amplitude modulation (QAM). In this case we can have any signal constellation we
choose. QAM waveforms are a combination of PAM and PSK. The basis functions are as for
PSK system. The QAM signal waveforms expresses as

s m (t ) = ∑ Am e jθ m g (t − mT )e j 2πfc t
m (7-1)
= Am g (t )[cos(2πf c t + θ m ) + sin (2πf c t + θ m )]

Where Am = (AI2 + AQ 2 )1/2. And AI and AQ are information bearing signal amplitudes of the
quadrature carriers g(t) is the signal pulse. In the other book reference the parameter Am.g(t) is
simplified as Am(t), which indicate a modulation amplitude. The parameter θm which indicate a
phase modulation, has a value:
θm = tan-1 (AQ /AI) (7-2)
We may choose any combination of M 1 -level PAM and M2 -level PSK to give an M=M 1M 2
level QAM signal constellation. Figure (7.1) shows some rectangular signal constellations for
different values of M. The optimum signal constellation will be that which requires the least
average power for a given minimum distance between signal points. For M=4 this will be when
the four points lie on a circle and the resulting signal constellation is the same as that for QPSK.

Tri Budi Santoso 95

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

For M=8 the optimal signal constellation consists of the points lying on two circles as indicated
in Figure (7.2).

Figure (7.1). Several signal space Figure (7.2) Example of PAM-PSK

For M=16 there are many more ways of building up the signal constellation however
rectangular signal constellations as shown in Figure (7.3a) have the distinct advantage that they
can be generate as two PAM signal impressed on the phase quadrature carriers and they are
easily demodulated. We say it as 16QAM by rectangular constellation or in some book call as
square constellation. Also for M=16 the average transmitter power required for the rectangular
constellation is only slightly worse than the best M-ary QAM signal constellation. The signal
output of 16QAM has general form as like Figure (7.3b).

Q V(t)
A3 +A3
+A2
A2
+A1
A1 I t

-A1

-A2
-A3

a) 16QAM Rectangular b) Signal output representative

Figure (7.3). The rectangular constellation of 16 QAM and signal output of 16QAM

The transmitter of 16QAM as the Figure (7.4). By assumption that signal input is a
sequence of 4 bits pair, the next process is serial to parallel conversion. Two MSB bits in Q

Tri Budi Santoso 96

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

channel, and the others are in I channel. The parallel bit information encoded by using Gray
code. The next process is mapping to 16QAM rectangular constellation (more detail in the
section 7.1.2). Each channel modulates the amplitude of carrier signal. I channel modulates the
sinusoid signal that has initial phase π/2 radiant (cos2πf ct), then called as in-phase channel. Q
channel modulates the sinusoid signal that has initial phase 0 radiant (sin2πf ct) and then called
as quadrature channel. The carrier-modulated signal are summing and transmitting.

Product
I channel
Gray

Constellation mapping
16QAM Rectangular
Code
cos(2πf ct )
Info S- P
Generator Conversion
π/2

Gray
Code Q channel
Product
Figure (7.4) 16QAM Transmitter

7.1.2. 16QAM Rectangular Constellation

In this section we will describe how to map the bits input to rectangular constellation. Four
ways mapping, will describe are:
- Natural binary code of 16QAM
- 2D gray code 16QAM
- 16QAM as sum of QPSK
- 16QAM as sum of DQPSK
Q Q
Q channel Q channel
1100 1101 1110 1111 1000 1001 1011 1010
+3 11 +3 10

1000 1001 1010 1011 1100 1101 1111 1110

+1 10 +1 11
I I
0 0
-1 01
0100 0101 0110 0111 0100 0101 0111 0110
-1 01

-3 00
0000 0001 0010 0011 -3 00 0000 0001 0011 0010

00 01 10 11 I channel 00 01 11 10
I channel

-3 -1 0 +1 +3 -3 -1 0 +1 +3

a) Natural binary Code of 16QAM b) 2D Gray Code of 16QAM

Figure (7.5) The 16 QAM Rectangular Constellation by Natural and 2D Gray Coded

Tri Budi Santoso 97

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• Natural binary code of 16QAM

In the natural binary code of 16QAM, the 2 bits pair at Q channel and I channel are
naturally coded. The2 bits pair input, naturally coded and Q channel representative are like in
the Table 1. The constellation is like Figure (7.5a). From this figure we see that between two
points nearest two bits different is possible. If receiver part makes a mistake in symbol recovery,
there will be one or two bit mistake.
If we have a sequence input: 0010, 1000,1111,and 0101. The 2 bits pair outputs of
modulator Q channel are 00, 10, 11, and 01. Q channel output is -3sin(2πf ct), +1sin(2πf ct),
+3sin(2πf ct), and -1sin(2πf ct). The 2 bits pair output of I channel are 10, 00, 11, and 01. I
channel output are -1cos(2πf ct), +3cos(2πf ct), -3cos(2πf ct), and +1cos(2πf ct). By using equation
(7-1) we will get the output of transmitter is 101/2 exp(2πf ct + 247.5o ), 101/2 exp(2πf ct + 157.5o ),
181/2 exp(2πf ct + 45o ), and 21/2exp(2πf ct + 225o ).

Table 7.1. Comparison of Natural and gray coded channel output

Bits pair Naturally coded Gray Coded
input
Q channel out I channel out Q channel out I channel out
Q I Q I Q I
put put put put
00 00 00 -3 sin(2πfc t) 00 -3 cos(2πfc t) 00 -3 sin(2πfc t) 00 -3 cos(2πfc t)
01 01 01 -1 sin (2πfc t) 01 -1 cos(2πfc t) 01 -1 sin(2πfc t) 10 -1 cos(2πfc t)
10 10 10 +1 sin (2πfc t) 10 +1 cos(2πfc t) 11 +1 sin(2πfc t) 11 +1 cos(2πfc t)
11 11 11 +3 sin (2πfc t) 11 +3 cos(2πfc t) 10 +3 sin(2πfc t) 10 +3 cos(2πfc t)

• 2D Gray Code 16QAM

In the 2 dimension (2D) gray code of 16QAM, the gray encoded data of channel Q and
channel I then mapped on to 16QAM Rectangular constellation. The 2-bit pair input, gray coded
and Q channel representative is like in the Table 1. The constellation is like Figure (1.5b). From
this figure we see that between two points nearest is only one bit different. If receiver part make
a mistake in symbol recovery, there will only one mistake.
If we have a sequence input: 0010, 1000,1111, and 0101. After gray coding process, the 2
bits pair outputs of modulator Q channel are 00, 11, 10, and 01. Q channel output is -3sin(2πf ct),
+1sin(2πf ct), +3sin(2πf ct), and -1sin(2πf ct). The 2 bits pair outputs of I channel are 11, 00, 10,
and 01. I channel outputs are +1cos(2πf ct), -3cos(2πf ct), +3cos(2πf ct), -1cos(2πf ct). By using
equation (7-1) we will get the output of transmitter is 101/2 exp(2πf ct + 247.5o ), 101/2 exp(2πf ct +
157.5o ), 181/2 exp(2πf ct + 45o ), and 21/2exp(2πf ct + 225o ).

Tri Budi Santoso 98

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• 16QAM as sum of QPSK

The signal constellation of 16QAM system as sum of QPSK is divided by four quadrants
as Figure (7.6). In this system the rule mapping of constellation point in the 16 QAM is
determine by the Q bits information:

- If the Q bits information is 00, bits information of I channel are placed same with the QPSK

system and the position at quadrant I.

- If the value of Q bits information are 01, I bits information are placed at quadrant II and

rotated π/2 radiant from their QPSK original position.

- If the value of Q bits information are 11, I bits information are placed at quadrant III and
rotated π radiant from their QPSK original position.
- If the value of Q bits information are 10, I bits information are placed at quadrant IV and
rotated -π/2 radiant from their QPSK original position.

If we have sequence input: 0010, 1000,1111,0101. The 2 bits pair outputs of Q channel are
00 with the signal point at quadrant I, 11 with the signal point at quadrant III, 10 with the signal
point at quadrant IV, and 01 with the signal point at quadrant II. The 2 bits pair outputs of I
channel are 11 not rotate from QPSK original position, 00 rotated π radiant from QPSK original
position, 10 rotated –π/2 radiant from original QPSK position, and 01 rotated π/2 radiant from
original QPSK position. The2 bits pair input, naturally coded and Q channel representative are
like in the Table 2.
Table 2. Bit input and channel output of 16QAM as sum of QPSK
Bits pair 16QAM as sum of QPSK
input
Q channel out I channel out
Q I Q I
put put
00 00 00 +3 sin(2πfct) 00 +3 cos(2πfct)
00 01 00 +3 sin(2πfct) 01 +1 cos(2πfct)
00 10 00 +1 sin(2πfct) 11 +1 cos(2πfct)
00 11 00 +1 sin(2πfct) 10 +3 cos(2πfct)
01 00 01 +3 sin(2πfct) 00 -3 cos(2πfct)
01 01 01 +1 sin(2πfct) 01 -3 cos(2πfct)
01 10 01 +1 sin(2πfct) 11 -1 cos(2πfct)
01 11 01 +3 sin(2πfct) 10 -1 cos(2πfct)
10 00 11 -3 sin(2πfct) 00 -3 cos(2πfct)
10 01 11 -3 sin(2πfct) 01 -1 cos(2πfct)
10 10 11 -1 sin(2πfct) 11 -1 cos(2πfct)
10 11 11 -1 sin(2πfct) 10 -3 cos(2πfct)
11 00 10 -3 sin(2πfct) 00 +3 cos(2πfct)

Tri Budi Santoso 99

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

11 01 10 -1 sin(2πfct) 01 +3 cos(2πfct)
11 10 10 -1 sin(2πfct) 11 +1 cos(2πfct)

I m info: 01 I I m info: 00
Q
00 10 m 01 00
+3
Re
01 11 11 10
+1
I
0
10 11 11 01
-1

00 01 10 00
-3
I m info: 11 I m info: 10

Figure (7.6). The 16QAM sum of QPSK

11 11 10 -3 sin(2πfct) 10 +1 cos(2πfct)
Based on the constellation signal as Figure (7.6) and Table 2, we will know that the outputs
of modulator Q channel are +1sin(2πf ct) -3sin(2πf ct), -3sin(2πf ct), and +1sin(2πf ct). The outputs
of modulator I channel are +1cos(2πf ct), -3cos(2πf ct), +1cos(2πf ct), and -1cos (2πf ct). By using
equation (1-1) we will get the output of transmitter is 21/2 exp(2πf ct + 45o ), 181/2 exp (2πf ct +
225o ), 21/2 exp (2πf ct - 67.5o ), and 21/2exp (2πf ct + 112.5o ).

• 16QAM as sum of DQPSK

The signal constellation of 16QAM system as sum of DQPSK is similar with the
constellation of 16QAM system as sum of QPSK. But in this system the rule of rotation is based
DQPSK system. The signal point at present time is a function of the signal point at the previous
time. In this system have two steps the transition point. First step is transition point for Q
channel, the second step is transition for I channel.
Q channel output

s1 s1
Q channel input

Differential
Encoder EXOR-1
QPSK
s1 s1

EXOR-3
I channel output

s2 s2
I channel input

EXOR-2
AND
s2 s2
EXOR-4

Tri Budi Santoso 100

Figure (7.7) Logic circuit mapping of 16QAM as sum of DQPSK
Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

By using the rule transition of DQPSK system, mapping of constellation point in the 16
QAM of Q bits information is:
- If the Q bits information are 00, there is no transition.
- If the value of Q bits information is 01, is rotated π/2 radiant from previous position.
- If the value of Q bits information is 11, is rotated π radiant from previous position.
- If the value of Q bits information are 10, is rotated - π /2 radiant from previous position.
For initial condition is made by assumption that 2 bits at Q channel are s11_0 = 0 and s11_0 = 00.

The transition point for I channel based on the logic circuit at Figure (7.7). The output of I
channel is a function of the output from Differential Encoding of Q channel and bits information
input of I channel.
If we have a sequence input: 0010, 1000,1111,0101. The 2 bits pair inputs of Q channel after
gray coding are 00, 11, 10, and 01. The Q channel output for after differential-encoding process
is 00, 11, 01, and 11. I channel inputs after gray coding are 10, 00, 11, and 01. Based on Figure
(1.7) we will know that the output of I channel are 10, 00, 11, and 01.

Im
Q channel
Im
00 10 01 00
+3

1 01 11 11 10
3 3
+1 1

0 Re

Re
10 11 11 01
-1

00 01 10 00
-3 2 4
2 4

I channel

a) Rotation process based on DQPSK b) Transition process at 16QAM sum of DQPSK

Tri Budi Santoso Figure (7.8). The 16QAM 101

as sum of DQPSK
Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

Based on the constellation signa l as Figure (7.8b), the outputs of modulator Q channel are
+1sin(2πf ct), -3sin(2πf ct), +1sin(2πf ct), and -3sin(2πf ct). The outputs of modulator I channel are
+3cos(2πf ct), -3cos(2πf ct), -1cos(2πf ct), and -1cos (2πf ct). By using equation (1-1) we will get
the output of transmitter is 101/2 exp (2πf ct + 22.5o ), 181/2 exp (2πf ct + 225o ), 21/2 exp (2πf ct +
135o ), and 101/2 exp (2πf ct + 247.5o ).

7.1.3. The Receiver of 16QAM

The receiver of 16QAM is similar with the QPSK system, but in this system each channel
content of two bits information. Generally the block diagram of 16QAM receiver is like Figure
(7.9). Same with transmitter part, the different among them is at de-mapping point of
constellation signal. At this section we make an assumption that the local carrier oscillator
exactly synchronized, and phase and frequency of local carrier of receiver part is same with
which used at transmitter part.

Product
I channel
LPF Gray
Decode
Local carrier ?
16QAM Detection and

cos (2πf ct)

De-mapping

P -S Information
Conversion recover
Received signal
r(t) π/2

Gray
LPF Decode
Q channel
Product

Figure (7.9) The 16QAM Receiver

After filtering process by using LPF, the PAM signal of each channel is detected to make a
decision of their levels. The next process is de-mapping, and these stages depend on the
mapping system that used by transmitter part. If the mapping system in transmitter part using
natural binary code, the de-mapping at receiver must use natural binary decode, so with 2 D
gray code, sum of QPSK and sum of DQPSK de -mapping system.

• Natural Binary Decode

In this technique Gray Decode part is not used. The PAM signal output from LPF is

Tri Budi Santoso 102

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

decode by using Natural binary code. If the received signal is 101/2 exp(2πf ct + 247.5o ),
101/2 exp(2πf ct + 157.5o ), 181/2 exp(2πf ct + 45o ), and 21/2exp(2πf ct + 225o ). After demodulation and
LPF process the output from Q channel output are -3, +1, +3, and -1. Normalizing the level will
change the Q channel output be 0, 2, 3, and 1. By using binary decode we will get the output as
00, 10, 11 and 01. At I channel demodulation and LPF process give the output at I channel as
are +1, -3, +3, and -1. Normalizing the level will change the Q channel output be 2, 0, 3, and 1.
By using binary decode we will get the output as 10, 00, 11 and 01. By using parallel to serial
conversion the output of bit recovery is 0010, 1000,1111, and 0101.

• 2 D gray Decode
In this technique similar with Natural binary code, but in this technique use Gray Decode.
After the process in demodulator and LPF, the PAM signal output is decoded by using gray
decode. If the received signal 101/2 exp(2πf ct + 247.5o ), 101/2 exp(2πf ct + 157.5o ), 181/2 exp(2πf ct +
45o ), and 21/2 exp(2πf ct + 225o ).
After demodulation and LPF process the output from Q channel output is -3, +1, +3, and -
1. Normalizing the level will change the Q channel output be 0, 2, 3, and 1. De-mapping process
by using 2D gray will give the output at Q channel as 00, 11, 10 and 01. By using gray decode
we will get the output will be 00, 10, 11, and 01. At I channel demodulation and LPF process
give the output at I channel as are +1, -3, +3, and –1. Normalizing the level will change I
channel output be 2, 0, 3, and 1. De-mapping process by using 2D gray will give the output at I
channel as 11, 00, 10 and 01. By using gray decode we will get the output as 10, 00, 11 and 01.
After parallel to serial conversion the output of bit recovery are 0010, 1000,1111, and 0101.

• Sum of QPSK De -mapping

In this technique after demodulator and LPF process, de-mapping signal is based on QPSK
de-mapping. If the received signal is 21/2 exp(2πf ct + 45o ), 181/2 exp (2πf ct + 225o ), 21/2 exp (2πf ct -
67.5o ), and 21/2 exp (2πf ct + 112.5o ). After demodulation and LPF at channel, the output of Q
channel is +1, -3, -3, and +1. At I channel the output is +1, -3, +1, and -1.
Based on the Figure (7.6) we see that these signal are located at quadrant I, III, IV, and II
respectively. Based on this position we know that the Q channel output are 00, 11, 10, and 01..
By using gray decode we will get the output of Q channel are 00, 10, 11, and 01. From
constella tion signal at Figure (7.6) we see that I channel has bit output as 11, 00, 10, and 01. By
using gray decode I channel output will be 10, 00, 11, and 01. By using parallel to serial
conversion the output of bit recovery are 0010, 1000,1111, and 0101.

Tri Budi Santoso 103

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• Sum of DQPSK De-mapping

In this technique after demodulator and LPF process, de-mapping signal is based on DQPSK
de-mapping. If the received signal is 101/2 exp (2πf ct + 22.5o ), 181/2exp (2πf ct + 225o ), 21/2 exp
(2πf ct + 135o ), and 101/2 exp (2πf ct + 247.5o ). After demodulation and LPF at channel, the output
of Q channel is +1, -3, +1, and -3. At I channel the output is +3, -3, -1, and -1. Based on the
Figure (7.8) we see that the bits of Q channel are 00, 11, 01, and 11. And the bits of I channel
are 10, 00, 11, and 01. By using differential decoding process as Figure (7.10) and bit 00 as
initial condition we will get the Q channel is 00, 11, 10, and 01. After gray decodes process it
will be 00, 10, 11, and 01. At I channel we will get the bit output as 11, 00, 10, and 01. After
gray decodes process it will be 10, 00, 11, and 01. After parallel to serial conversion the output
of bit recovery are 0010, 1000,1111, and 0101.

Q channel output
s1 s1
Q channel input

Differential EXOR-1
Decoder
QPSK
s1 s1

EXOR-3

s2

I channel output
s2
I channel input

EXOR-2
AND
s2 s2
EXOR-4

Figure (7.10) Logic circuit de-mapping of 16QAM as sum of DQPSK

7.1.3. 16QAM Transmission in the AWGN Channel

The 16QAM-transmission system is similar with QPSK transmission, but in 16QAM
transmission each channel contains of 2 bits or the other hand that energy average each channel
of 16QAM l is twice compare than in the QPSK channel. The simplification for it can be
described as Figure (7.11).

16QAM 16QAM Receiver

Transmitter r(t) = s(t) + n(t)

White Gaussian Noise

n(t) = n re(t) + n im(t)

Figure (7.11) 16QAM transmission through AWGN channel

Tri Budi Santoso 104

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

In the AWGN channel, each channel of 16QAM (In-phase and Quadrature or in the some
books reference is called as Real and Imaginary) will disturbed by two independent additive
noise which Gaussian random.
The signal transmission of 16QAM system is:
s(t) = Re(t) + Im(t)
Where Re(t) is represent the I channel of 16QAM and Im(t) is represent the Q channel of
16QAM. The noise generate from AWGN channel is:
n(t) = nre(t) + nim (t)
The addition process in the channel will make the received signal at receiver part as
r(t) = s(t) + n(t)
= [Re(t) + nre(t)] + [Im(t) + nim (t)] (1-5)
Im
δ

A
A r2 r1 Re

r3
A

Figure (7.12). Three groups signal

of 16QAM rectangular

To evaluate the performance of 16QAM in the AWGN channel, it must consider its
constellation signal. By modification Figure (7.2) we will get three groups with different energy
average. The distance between two nearest signals is δ, in this case is represent the energy
different between two nearest point signals.

- For group A 1 with r1 = 2δ 2 , the energy average each signal equivalent with r1 2 = δ 2 /2.

- For group A2 with r2 = 5 2δ , the energy average each signal equivalent with r2 2 = 5δ2 /2.

- For group A2 with r2 = 2 (3/2) δ 2 , the energy average each signal equivalent with r2 2 =

Tri Budi Santoso 105

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

9δ 2 /2.
The total energy for 16-signal point of 16QAM rectangular constellation is:
Etot = (4 x δ2 /2) + (8 x 5δ 2 /2) + (4 x 9δ 2 /2). (7-6)

The average energy of each signal is:

Es = [(4 x δ 2 /2) + (8 x 5δ 2 /2) + (4 x 9δ 2 /2)]/16 = 5 δ 2 /2 (7-7)
The AWGN channel gives noise, which has a variance σ2 . Average energy to noise ratio of the
received signal at receiver is:
Es /No = (5/2) δ2 /σ2 (7-8)
And bit energy to noise ratio is:
Eb /No = (5/8) δ2 /σ2 (7-9)

Performance of 16QAM system in the AWGN channel in the case of symbol error can be
evaluated as a function of error function of Energy to noise ratio, and expressed as:

M −1    
Ps = 2 erfc 
3 Es  = 2 16 − 1 erfc  3 4 Eb 
M  2 (M − 1) No 
 16 
 2(16 − 1) No 

3  2 E b 
= erfc  (7-10)
2  5 No 


By include the value in equation (7-9) we will get the value of symbol error probability as:

3  25δ2 
Ps = erfc  (7-11)
2  5 8σ 2 
 
And the bit error probability is

3  2 5δ2 
Pb = erfc  (7-12)
8  58σ2 
 
The symbol error probability performance of 16QAM system in the AWGN channel as a
function of energy per bit to noise ratio can be described as Figure (7.13).

Tri Budi Santoso 106

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

1
'QPSKQAMt.txt'

0.1

0.01
Bit error probability, Pb

0.001

0.0001

1e-05

1e-06
6 8 10 12 14
Eb/No in dB

Figure (7.13). Performance of 16 QAM rectangular constellation in the AWWGN channel

7.1.4. Carrier-phase Synchronization of 16QAM System

For the real system the frequency and initial phase of transmitter part is unknown by the
receiver part. At the receiver part must be generated a local carrier with initial phase estimated.
The system will work after the frequency and phase of receiver part has been synchronized to
the frequency and phase of transmitter part. The process synchronization then called as carrier-
phase recovery. In this example we use Costas Loop as carrier-phase recovery for
synchronization process. For it we denote the frequency carrier at transmitter as f cT, and local
carrier frequency generated by receiver as f cR. The initial phase of transmitter signal as θi , and
the phase estimate at receiver part as θest.
Carrier-phase recovery by using Costas Loop for 16QAM is a combination from two parallel
carrier-phase recoveries, which we have applied it for QPSK system. But in this case we must
consider the amplitude of received signal. General descriptions of Costas loop carrier recovery
for 16QAM in the base band term as the Figure (7.14). There is two parts in the carrier recovery
for 16 QAM, carrier-phase detector and amplitude detector.

Tri Budi Santoso 107

 Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

Amplitude
Product1 Detector
LPF
Acos(2pfcTt+φi) Addition1 Product5
cos(2πfcRt+φest)
I
ch LPF
Product2
Product7 S/H
Addition2
sin(2πf cRt+φest)
Product6
Product3
LPF
+ Substraction2
Asin(2pfcTt+φi)
Q cos(2πfcRt+φest) −
ch
Substraction1
LPF
Product4 Amplitude Detector
sin(2πf cRt+φest)

2πf cRt + φ est + K ∫ sin [4(2π∆ft + ∆φ )] VCO

0

Figure (7.14) Simplification of carrier recovery for QPSK system

• Carrier-phase detector
In QPSK system the transmission signal is content of Real (Re) and Imaginary (Im) part:
s(t) = A Recos(2πf cTt +φ i ) + AImcos(2πf cTt +φi ).
In this case the average value of ARe and AIm is 1. The synchronization process by using Costas
Loop is as Figure (7.14). In synchronization process for 16QAM each channel is multiply by
cos(2πf cRt +φ est ) and sin(2πf cRt +φ est )

• The output of Product 1 is:

yp1 (t) = cos(2πf cTt + φ i ) cos (2πf cRt + φ est) + n(t) Recos(2πf cRt + φ est )
= (1/2)cos{2π( f cT + f cR)t + (φi + φest)} + (1/2)cos{2π( f cT - f cR)t + (φ i - φ est)}
+ nRe(t)cos (2πf cRt + φ est )
The low pass filter process will give the output as:
yLPF1 (t) = (1/2)cos{2π( f cT - f cR)t + (φi - φ est )}= (1/2)cos(2π ∆f t + ∆φ t )

Tri Budi Santoso 108

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• The output of Product 2 is:

yp1 (t) = cos(2πf cTt + φ i ) (sin(2πf cRt + φ est)) + n(t) Resin(2πf cRt + φ est)
= (1/2)sin{2π( f cT + f cR)t + (φ i + φ est )} - (1/2)sin{2π( f cT - f cR)t + (φi - φest)}
+ nRe(t)sin(2πf cRt + φ est)
The low pass filter process will give the output as:
yLPF2 (t) = (-1/2)sin{2π( f cT - f cR)t + (φi - φest)}= (-1/2)sin(2π ∆f t + ∆φ t )

• The output of Product 3 is:

yp3 (t) = sin(2πf cTt + φi ) cos (2πf cRt + φest) + n(t) Imcos(2πf cRt + φ est)
= (-1/2)sin{2π( f cT + f cR)t + (φ i + φ est )} + (1/2)sin{2π( f cT - f cR)t + (φ i - φ est)}
+ nI m(t)cos (2πf cRt + φ est )
The low pass filter process will give the output as:
yLPF3 (t) = (1/2)sin{2π( f cT - f cR)t + (φ i - φ est)}= (1/2)sin(2π ∆f t + ∆φt )

• The output of Product 4 is:

yp4 (t) = sin(2πf cTt + φi ) sin(2πf cRt + φ est) + n(t) Imsin(2πf cRt + φ est )
= -(1/2)cos{2π( f cT + f cR)t + (φi + φest)} + (1/2)cos{2π( f cT - f cR)t + (φ i - φ est)}
+ nI m(t)sin(2πf cRt + φ est)

The low pass filter process will give the output as:
yLPF4 (t) = (1/2)cos{2π( f cT - f cR)t + (φi - φ est )}= (1/2)cos(2π ∆f t + ∆φ t )

The next steps are

• The output of Addition1 is:
add1(t) = (1/2)cos(2π ∆f t + ∆φ t )+(1/2)cos(2π ∆f t + ∆φt ) = cos(2π ∆f t + ∆φt )

• The output of Substraction1 is:

subst1(t) = (1/2)sin(2π ∆f t + ∆φt )-(-1/2)sin(2π ∆f t + ∆φt ) = sin(2π ∆f t + ∆φt )

• The output of Product5 is:

yp5 (t) = cos(2π ∆f t + ∆φt ) sin(2π ∆f t + ∆φ t) = (1/2)sin{2(2π ∆f t + ∆φ t )}
• The output of addition2 is:
add2(t) = cos(2π ∆f t + ∆φt ) + sin(2π ∆f t + ∆φt )

Tri Budi Santoso 109

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• The output of Subtraction2 is:

Sub2 = cos(2π ∆f t + ∆φt ) - sin(2π ∆f t + ∆φt )
• The output of Product6 is:
yp6 (t) = [cos(2π ∆f t + ∆φt ) + sin(2π ∆f t + ∆φt )][ cos(2π ∆f t + ∆φt ) - sin(2π ∆f t + ∆φ t )]
= cos 2(2π ∆f t + ∆φt ) – sin2 (2π ∆f t + ∆φt )
= cos{2(2π ∆f t + ∆φt )}

• The output of Product7 or the error loop signal is:

e(t) = (1/2) sin{2(2π ∆f t + ∆φt )}cos{2(2π ∆f t + ∆φ t )}
=(1/4)[2 sin{2(2π ∆f t + ∆φt )}cos{2(2π ∆f t + ∆φ t)}]
=(1/4) sin{4(2π ∆f t + ∆φt )} (7-13)

It is known as error signal into the loop filter of QPSK carrier-phase recovery, and used to drive
the VCO.
The new output phase of VCO is given as
t

2πf cR t + φ est + ∫ K c e(t ) dt

0

or (7-14)
t

2πf cR t + φ est + K ∫ sin (4( 2π∆ft + ∆φ ) )dt where is K = Kc/4.

0

The loop phase error and its time derivative is

 t

θ e ( t ) = (2πf cT t + φ i ) −  2πf cR t + φ est + K ∫ sin (4( 2π∆ft + ∆φ ) )dt 

 0 
d (θ e ( t ) )
= (2πf cT − 2πf cR ) − K sin (4( 2π∆ft + ∆φ ) )
dt
If f cT = f cR and letting K =1, the time derivative of the loop phase error is given by
d (θ e ( t ) )
= − sin (4∆φ ) (7-15)
dt
The null value will be happen at 0,π/2, π, 3π/2 radiant, and it’s known as four phase ambiguity
of phase-carrier recovery at 16QAM system.

Tri Budi Santoso 110

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• Amplitude detector
Consider the constellation signal in Figure (7.12). It can be modified as Figure (7.15). In its
figure, there is:
- Group A consists of signals that have amplitude value: amplitude< or amplitude > 14.
- Group B consists of signals that have amplitude value: 6 < amplitude < 14.

(I’)2 + (Q’)2 < 6

Group A
or (I’)2 + (Q’)2 > 14

Group B
6 < (I’)2 + (Q’)2 < 14

Figure (7.15). Two groups signal

If the received signal is group A, the group detector drive sample and hold (S/H) to make an
action for the phase detected by carrier-phase recovery is used and hold its value. In the case
where the received signal is group B, the group detector drive S/H to make an action that the
phase detected by carrier-phase recovery is unavailable and used the previous phase detected
from previous carrier-phase recovery process.
The carrier-phase recovery for 16QAM is only capable to detect the phase of received signal
in the group A, but this system can’t detect the phase of the received signal in the group B.

Tri Budi Santoso 111

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

7.2. Session 2: Simulation System

We will tray to develop the 16QAM simulation in the base band system and using 2D Gray to
develop a rectangular constellation. By using Costas Loop as phase-carrier recovery at the
receiver part, we will tray to understand the performances of 16QAM system through the
AWGN channel. The description of 16QAM for base band simulation as in the Figure (7.16).

I_Q map Noise addition

Info I_Q Gray Symbol

generate mapping Code transmit

Im channel

Re channel
Error rate Error
count detect
AWGN
generation
I_Q De-map

Info I_IQ de- Gray de- Symbol

received map Code received

Synchronization
∆φ

Figure (7.16) Block diagram of 16QAM Base band simulation

7.2.1. Algorithm
The simulation of 16QAM through AWGN channel involving several functions as follows:

• Info Generator
Bit information generated randomly by info generator function. The random number
generated is set value 0,1,2 ~ 15, as the integer number to represent PAM signal. By using a
command (int)rand%16; of the Microsoft C, it will generate a number from 0 ~15 uniform
randomly.

• I - Q Channel Mapping
This function consists of three parts in the Figure (7.16): I_Q Map, Gray Code and Symbol
Transmit.

• The I_Q Map. In this program we denote I channel as Real (Re) channel and Q channel

Tri Budi Santoso 112

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

we denote as Imaginary (Im) channel. The information will be mapped into Re and Im
channel. The decimal value is converted to binary, the two MSB is mapped into Im
channel, and the two LSB is mapped into Re channel. Based on the Figure (7.17a) we
will understand that if the information value is 0 the two bits pair generates is 0000, the
Im channel will have the value 00, and Re channel will have the value 00. If the
information value is 15 the two bits pair generates is 1111, the Im channel will have the
value 11, and Re channel will have the value 11. It can do by using this statement:
Im_info1[i]=info[i]/8;
Im_info2[i]=(info[i]-Im_info1[i]*8)/4;
Re_info1[i]=(info[i]-Im_info1[i]*8-Im_info2[i]*4)/2;
Re_info2[i]=(info[i]-Im_info1[i]*8-Im_info2[i]*4-Re_info1[i]*2);

• The Gray Code. Re channel and Im channel info has natural bit value, based on Table 7.1
and the logic as like in the Figure (7.16b) we able to make Gray code logic. If the value
of Re channel is 00, it will coded as 00, if the value is 01 it will coded as 01, if the value
is 10 it will coded as 11, and if the value is 11 it will coded as 10. It is same for Im
channel. From the above case the value 0000 will be 0000 and the value 1111 will be
1010.
• The Symbol Transmit. By using the properties in the Table 7.1 wee can develop the
symbol representative for each channel. This part it will convert the value 00 as –3, the
value 01 as –1, the value 11 as +1, the value 10 as +3.

Im channel

23 Natural Gray
Im_info1
Im_ info1

coded
22 Im_ info2
Info
EXOR
21 Re_ info1 Re_ info1

coded
20 Re_ info2
EXOR
Re channel
a) I_Q Map b) Gray Code

Figure (7.17) Description of channel mapping

• Synchronization
In this function we must make sure that receiver part understand exactly the phase of the
received signal from transmitter. Unfortunately in the real system, local carrier frequency and its

Tri Budi Santoso 113

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

initial phase does not same with the transmitter part. The Costas loop, as carrier recovery is uses
to correct the phase different.
As we know that in the synchronization process by using Costas loop for 16QAM system
there two phase ambiguity among 0, π/2, π, and 3π/2 radiant. Using this property the algorithm
looks for one of two possible phases. When it matched we must resolve by de-rotate the phase.

• AWGN
To generate AWGN channel we must do it by using two steps: uniform random generate and
shifting from uniform to Gaussian distribution.
• First step is generates two sequences of uniform random generator, for this purpose we
can set variable x1 and x2. Both of them have double data type. These two variables
have a value between 0 and 1. To do it we can generate the value x1 and x2 from 0 until
default value of RAND_MAX or 32767.00. Then we divide x1 and x2 by
RAND_MAX.
• Second step do by using Box-Muller method. Two variable x1 and x2, which have the
uniform distribution, shift by using formulation:
re = (σ2 lnx1)1/2 cos(2*π*x2) ;real part
im = (σ2 lnx1)1/2 sin(2*π*x2) ;imaginary part.

• Noise Addition
The noise generated from the AWGN function are +1 Volt

real and Imaginary part, and independent each one t

another. In this function we add the Re channel with

real noise and the Im channel with imaginary noise. In a) Decreasing level
+1 Volt
some books Re channel from 16QAM transmitter is
called as real channel, and Im channel is called as t

imaginary channel. AWGN will destroy the amplitude

b) Increasing level
of signal transmission, and will depredate the level of
Figure (7.18). The effect AWGN at the level of
signal. If the value of AWGN is negative it will make
antipodal signal transmission
the level of signal transmission is down, other wise if
the AWGN value is positive is will make the level of signal transmission will up.
The effect of AWGN will change the level of transmitted signal as like in the Figure (7.18).

Tri Budi Santoso 114

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• Detection
This function represents the Symbol Received part. Based on + 3.00
the several levels as a threshold, the decision is made:
if signal level < -2.0 the decision is –3.0 + 1.00

if -2.0 < signal level < 0.0 the decision is -1.0 0.00
if 0.0 < signal level < 2.0 the decision is +1.0 − 1.00
if signal level > 2.0 the decision is +3.0

− 3.00
This rule for both of Re channel and Im channel signal received,
Figure (7.19) Decision
and work independently. symbol for received signal

• Rx De-map
This function consists of two parts of block diagram in the Figure (7.20).
• Gray de-Code. First step in this block is converts from antipodal signal to binary signal.
By using the mathematical expression opposite with we has used in the symbol
transmission. Second step is converts from gray form to natural binary from. By using
the XOR logic we get the natural binary result as like Figure (7.20a).
• I_Q De-map. The purpose of this part is get the symbol value by convert the two pair
bit (binary) decimal value. By multiply the Im channel by 23 and 22 and Re channel by
21 and 20 . The next step is addition the multiplication result and we will get the decimal
information value.

Gray Im channel
Natural
Im_decod1 23

Im_ decod2 Info

22
EXOR Recover

Re_ decod1 21

Re_ decod2 20
EXOR Re channel

b) Gray Decode b) I_Q De-Map

Figure (7.20) Description of Rx De-mapping

Tri Budi Santoso 115

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

• Error Detect
By using the output from detection function and compare bit by bit between input signal of
symbol transmit to output of symbol receive. If the value between both parts is the error didn’t
happen, other wise if these value is not same, the error was happening. By combine the result of
these result we will get the information of symbol error detection. If one bit or more error
happened, the symbol error detect indicate the value 1, if no error both of channel there is no
error happened and the value of symbol error detect indicate the value 0.

• Error Rate
In this part we sum all of error detected that happened during transmission process. And the
result from this summation subtracted by the total bit transmitted. The result is bit error rate. If
we compare the summation of symbol error detected to total symbol transmitted, the result is
symbol error rate. This is the real performance resulted by our transmission system.

♦ Pb_theory
In this part we calculate the bit error probability of 16QAM system as a function of energy
per bit to noise ratio. The output is theoretical value of error probability performance of
16QAM system. From it we get the information of bit error probability and symbol error
probability.

7.2.2. The Listing Program

To develop a program in order the user understand it easily we must set the symbol which
represent the parameters in theory:
N = total symbol transmit.
info= information generate by transmitter part.
Im_info1= Imaginary channel 1 information generated at transmitter part
Im_info2= Imaginary channel 2 information generated at transmitter part
Re_info1= Real channel 1 information generated at transmitter part
Re_info2= Real channel 2 information generated at transmitter part
Re_enc1= Real channel 1 information encoded at transmitter part
Re_enc2= Real channel 2 information encoded at transmitter part
Im_enc1= Imaginary channel 1 information encoded at transmitter part
Im_enc2= Imaginary channel 2 information encoded at transmitter part
pi = π radiant.

Tri Budi Santoso 116

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

Re_Tx = symbol generate by transmitter part.

Im_Tx = symbol generate by transmitter part.
pha_Tx = phase generate by transmitter part
phase = phase error of synchronization process
Ave_Pha_err = average phase error in the all the synchronization process
Re_AWGN = real part of noise AWGN
Im_AWGN = imaginary part of noise AWGN
var = noise varians
Re_Rx = Real channel symbol received at receiver part
Im_Rx = Imaginary channel symbol received at receiver part
pha_receiv = phase receive after transmission
Re_1_Rx = Real channel 1 information recovered at receiver part
Re_2_Rx = Real channel 2 information recovered at receiver part
Im_1_Rx = Imaginary channel 1 information recovered at receiver part
Im_2_Rx = Imaginary channel 2 information recovered at receiver part
SNR = Signal to Noise Ratio
BER = bit error rate
Pb = probability of bit error of 16QAM system
Ps = probability of symbol error of 16QAM system

The listing program of 16QAM system is as follows.

//Program of 16QAM
//editing by Tri Budi Santoso
//Suzuki Laboratory
//Mobile Communication Group, Tokyo Institute of Technology, Japan

#include<stdio.h>
#include<math.h>
#include<stdlib.h>

#define N 1000000

int nn=N-1,i;
int info[N],Im_info1[N],Im_info2[N],Re_info1[N],Re_info2[N];
int Re_enc1[N],Re_enc2[N],Im_enc1[N],Im_enc2[N];
double Re_Tx[N],Im_Tx[N],pha_Tx[N],Amp[N];
double Im_Rx[N],Re_Rx[N];
double phase[N],Ave_Pha_err,Pha_err,deg_error,pha_i;
int info_Rx[N],Im_1_Rx[N],Im_2_Rx[N],Re_1_Rx[N],Re_2_Rx[N];
double SNR,Ps,Pb,BER,SER,pi=acos(-1.0),var=0.2;
double Re_AWGN[N],Im_AWGN[N];

Tri Budi Santoso 117

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

void info_generate(int info[],int Im_info1[],int Im_info2[],int Re_info1[],int Re_info2[]);

void encoding(int Im_enc1[],int Im_enc2[],int Re_enc1[],int Re_enc2[]);
void mapping_transmit(double Re_Tx[],double Im_Tx[],double pha_Tx[],double Amp[]);
void costas(double phase[],double *pha_er);
void synchronization(double pha_i,double *deg_er);
void AWGN_channel(double Re_AWGN[],double Im_AWGN[]);
void detection(double Re_Rx[],double Im_Rx[]);
void demapping_recover(int Im_1_Rx[],int Im_2_Rx[],int Re_1_Rx[],int Re_2_Rx[]);
void error_detect(double *ber,double *ser);
void signal_noise(double *snr);
void error_prob(double Ave_Pha_err,double *ps, double *pb);

void main()
{
FILE *fdata;
char filename[]="natural.txt";
fdata = fopen(filename,"a");

info_generate(info,Im_info1,Im_info2,Re_info1,Re_info2);
encoding(Im_enc1,Im_enc2,Re_enc1,Re_enc2);
mapping_transmit(Re_Tx,Im_Tx,pha_Tx,Amp);
costas(phase,&Ave_Pha_err);printf("\nAve_Pha_err:%f",Ave_Pha_err);
AWGN_channel(Re_AWGN,Im_AWGN);
detection(Re_Rx,Im_Rx);
demapping_recover(Im_1_Rx,Im_2_Rx,Re_1_Rx,Re_2_Rx);
error_detect(&BER,&SER);
signal_noise(&SNR);
error_prob(Ave_Pha_err,&Ps,&Pb);

printf("\n SNR:%4.2f Pb:%7.6f BER:%7.6f Ps:%7.6f SER:%7.6f",SNR,Pb,BER,Ps,SER);

fprintf(fdata,"%4.2f %7.6f %7.6f",SNR,BER,SER);
fclose(fdata);
}

void info_generate(int info[],int Im_info1[],int Im_info2[],int Re_info1[],int Re_info2[])

{
for (i=1;i<=nn;++i)
{
info[i]=(int)rand()%15;
Im_info1[i]=info[i]/8;
Im_info2[i]=(info[i]-Im_info1[i]*8)/4;
Re_info1[i]=(info[i]-Im_info1[i]*8-Im_info2[i]*4)/2;
Re_info2[i]=(info[i]-Im_info1[i]*8-Im_info2[i]*4-Re_info1[i]*2);
//printf("\ninfo[%d]:%d \tIm_info1[i]:%d Im_info2[i]:%d Re_info1[i]:%d
Re_info2[i]:%d",i,info[i],Im_info1[i],Im_info2[i],Re_info1[i],Re_info2[i]);
}
}

void encoding(int Im_enc1[],int Im_enc2[],int Re_enc1[],int Re_enc2[])

{
for (i=1;i<=nn;++i)
{
Im_enc1[i]=Im_info1[i];
Im_enc2[i]=(Im_info1[i] + Im_info2[i])%2;
Re_enc1[i]=Re_info1[i];
Re_enc2[i]=(Re_info1[i] + Re_info2[i])%2;
//printf("\nIm_enc1[i]:%d Im_enc2[i]:%d Re_enc1[i]:%d

Tri Budi Santoso 118

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

Re_enc2[i]:%d",Im_enc1[i],Im_enc2[i],Re_enc1[i],Re_enc2[i]);
}
}

void mapping_transmit(double Re_Tx[],double Im_Tx[],double pha_Tx[],double Amp[])

{
for (i=1;i<=nn;++i)
{
Im_Tx[i]=(Im_info1[i]*2.0+Im_info2[i]*1.0)*2.0-3.0;
Re_Tx[i]=(Re_info1[i]*2.0+Re_info2[i]*1.0)*2.0-3.0;
Amp[i] = Re_Tx[i]*Re_Tx[i] + Im_Tx[i]*Im_Tx[i];
pha_Tx[i]=atan(Im_Tx[i]/Re_Tx[i]);
if (Re_Tx[i]>0.0 && Im_Tx[i]>0.0)
{pha_Tx[i]=pha_Tx[i]/pi*180;}
else if (Re_Tx[i]>0.0 && Im_Tx[i]<0.0)
{pha_Tx[i]=360.0 + pha_Tx[i]/pi*180;}
else if (Re_Tx[i]<0.0 && Im_Tx[i]>0.0)
{pha_Tx[i]=180.0 - pha_Tx[i]/pi*180;}
else if (Re_Tx[i]<0.0 && Im_Tx[i]<0.0)
{pha_Tx[i]=180.0 + pha_Tx[i]/pi*180;}
else
printf("");
// printf("\nIm_Tx[%d]:%4.2f Re_Tx[%d]:%4.2f Amp[%d]:%4.2f",i,Im_Tx[i],i,Re_Tx[i],i,Amp[i]);
}
}

void costas(double phase[],double *pha_er)

{
int input;
Pha_err=0.0;

for(i=1;i<=nn;++i)
{
pha_i=pha_Tx[i];
synchronization(pha_i,&deg_error);
phase[i]=deg_error;
if(Amp[i]<6.0)
{input=1;}
else if(Amp[i]>=6.0 && Amp[i]<=14.0)
{input=0;}
else if(Amp[i]>14.0)
{input=1;}
else
printf("");

if(input==0)
{phase[i]=phase[i-1];}
else if(input==1)
{phase[i]=phase[i];}
else
printf("");

//printf("\n phase[%d]:%f phase[%d]:%f",i-1,phase[i-1],i,phase[i]);

Pha_err += fabs(phase[i]);
}
Ave_Pha_err=Pha_err/nn;

Tri Budi Santoso 119

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

*pha_er=Ave_Pha_err; //printf("\nAve_Pha_err:%f",Ave_Pha_err);
}

void synchronization(double pha_i,double *deg_er)

{
int d=1,i=1,symbol_rate=16000,phase_uni;
double rad_error,fc=800e+6,T,delta_fc,phase_estim;
double VCO,K=1.0;

T=1.0/(double)symbol_rate;
delta_fc=1e -6 * fc *T;
do
{
phase_uni=(int)rand()%360;
phase_estim=(double)phase_uni;
deg_error=(delta_fc + pha_i - phase_estim);
//printf("\nphase_estim:%f deg_error:%f ",phase_estim,deg_error);
do
{
deg_error=(delta_fc + pha_i - phase_estim);
rad_error=deg_error/360*2*pi;
VCO=K*sin(4*rad_error);
phase_estim = phase_estim + VCO;
i++;
}while(i<=10);
d++;
} while(fabs(deg_error)>=4.0);
*deg_er=deg_error;
}

void AWGN_channel(double Re_AWGN[],double Im_AWGN[])

{
double x1,x2;
for(i=1;i<=nn;++i)
{
x1=(double)rand()/(RAND_MAX+1.0);
if (x1<=1e-38)
{x1=1e-38;}
x2=(double)rand()/(RAND_MAX+1.0);
Re_AWGN[i]=sqrt(-2.0*var*log(x1))*cos(2*pi*x2);
Im_AWGN[i]=sqrt(-2.0*var*log(x1))*sin(2*pi*x2);
//printf("Re_AWGN[i]:%f ,Re_AWGN[i]:%f \n",Re_AWGN[i],Im_AWGN[i]);
}
}

void detection(double Re_Rx[],double Im_Rx[])

{
double costas_effect=0.0;
for (i=1;i<=nn;++i)
{
costas_effect=cos(phase[i]/360*2*pi);
Im_Rx[i] = Im_Tx[i]*costas_effect + 1.75*Im_AWGN[i];
Re_Rx[i] = Re_Tx[i]*costas_effect + 1.75*Re_AWGN[i];
// printf("\nIm_Rx[i]:%f \tRe_Rx[i]:%f costas_effect:%f",Im_Rx[i],Re_Rx[i],costas_effect);
}
}

Tri Budi Santoso 120

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

void demapping_recover(int Im_1_Rx[],int Im_2_Rx[],int Re_1_Rx[],int Re_2_Rx[])

{
for (i=1;i<=nn;++i)
{
if(Re_Rx[i]<-2.0)
{Re_1_Rx[i]=0;Re_2_Rx[i]=0;}
else if(Re_Rx[i]>=-2.0 && Re_Rx[i]<0.0)
{Re_1_Rx[i]=0;Re_2_Rx[i]=1;}
else if(Re_Rx[i]>=0.0 && Re_Rx[i]<2.0)
{Re_1_Rx[i]=1;Re_2_Rx[i]=1;}
else if(Re_Rx[i]>2.0)
{Re_1_Rx[i]=1;Re_2_Rx[i]=0;}
else
printf("");

if(Im_Rx[i]<-2.0)
{Im_1_Rx[i]=0;Im_2_Rx[i]=0;}
else if(Im_Rx[i]>=-2.0 && Im_Rx[i]<0.0)
{Im_1_Rx[i]=0;Im_2_Rx[i]=1;}
else if(Im_Rx[i]>=0.0 && Im_Rx[i]<2.0)
{Im_1_Rx[i]=1;Im_2_Rx[i]=1;}
else if(Im_Rx[i]>2.0)
{Im_1_Rx[i]=1;Im_2_Rx[i]=0;}
else
printf("");
// printf("\nIm_1_Rx[i]:%d Im_2_Rx[i]:%d Re_1_Rx[i]:%d Re_2_Rx[i]:%d "
,Im_1_Rx[i],Im_2_Rx[i],Re_1_Rx[i],Re_2_Rx[i]);
}
}

void error_detect(double *ber,double *ser)

{
double BER1=0.0,BER2=0.0,BER3=0.0,BER4=0.0,BER_i,SER_i;
for (i=1;i<=nn;++i)
{
// BER_i=0.0;SER_i=0.0;
if(Re_1_Rx[i]==Re_enc1[i])
BER1=0.0;
else
BER1=1.0;
if(Re_2_Rx[i]==Re_enc2[i])
BER2=0.0;
else
BER2=1.0;
if(Im_1_Rx[i]==Im_enc1[i])
BER3=0.0;
else
BER3=1.0;
if(Im_2_Rx[i]==Im_enc2[i])
BER4=0.0;
else
BER4=1.0;

BER_i=(BER1 + BER2 + BER3 + BER4);

BER += BER_i;
if (BER_i>0.0)

Tri Budi Santoso 121

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

SER_i=1.0;
else
SER_i=0.0;
SER +=SER_i;
}
BER=BER/(4.0*nn);
SER=SER/nn;
*ber=BER;
*ser=SER;// printf("\nBER:%f SER:%f",BER,SER);
}

void signal_noise(double *snr)

{
double gama;
gama=(5./8.0)*2.0/var;
SNR=10*log10(gama);
*snr=SNR;
}

void error_prob(double Ave_Pha_err,double *ps, double *pb)

{
double gama,x,term_1,term_3,term_5;
gama=Ave_Pha_err*(5.0/8.0)/var;
x=sqrt(0.4*gama);
term_1=1/(2*pow(x,2));
term_3=(1*3)/(2*2*pow(x,4));
term_5=(1*3*5)/(2*2*2*pow(x,6));
Ps=3.0*(exp(-(x*x))/(x*sqrt(pi)))*(1-term_1+term_3-term_5);
*ps=Ps;
Pb=Ps/4.0;
*pb=Pb;
}

7. 2.3. Simulation Result

The simulation result for the 16QAM system is like in the Table 7.3.
Table 7.3. Simulation result of 16QAM in the AWGN channel

Eb/No Pb BER
6.20 0.01748865 0.03369800
7.96 0.00878699 0.01263550
10.97 0.00058334 0.00078775
11.43 0.00032042 0.00044950
11.94 0.00015218 0.00022350
12.52 0.00005879 0.00009800
13.19 0.00001669 0.00002800
13.98 0.00000290 0.00001175

From the Table 7.3 we can see that the different value between probability of error and bit
error rate (BER) as simulation result. As example for value of Eb/No = 6.02, probability of error
value is 0.01748865. But the simulation result is 0.03369800. In this case the bit error rate happen
is higher than the theoretical value (probability of error).

Tri Budi Santoso 122

Final Report of Telecommunication Engineering Training
Suzuki Laboratory, Mobile Communication Group
Tokyo Institute of Technology, Japan

For higher Eb/No, the value of bit error rate as simulation result is still higher than the value
of probability of error. The example is at Eb/No = 13.98 the value of probability of error is
0.00000290 and the value of bit error rate is 0.00001175.

1
'QPSKQAMt.txt'
'QPSKQAM.txt'

0.1

0.01
Bit error probability, Pb

0.001

0.0001

1e-05

1e-06
6 8 10 12 14
Eb/No in dB

Figure (7.13). The simulation result of 16 QAM rectangular constellation in the AWGN channel

Tri Budi Santoso 123