Probability of Error 1

©Dr. James S. Kang

Professor
ECE Department
Cal Poly Pomona

1
 Probability of Error for M-ary PAM
• N = 1 (1D)
• d = minimum distance between the signal vectors.
• The probability of correct decision is P(C)=(2/M)(1-p)+[[(M-2)/M](1-2p).
-(M-1)d/2 … -3d/2 -d/2 d/2 3d/2 … (M-1)d/2

• The probability of error is

2 𝑀𝑀−2 𝑀𝑀−1 1
• 𝑃𝑃 𝐸𝐸 = 1 − 𝑃𝑃 𝐶𝐶 = 𝑝𝑝 + 2𝑝𝑝 = 2 𝑝𝑝 =2 1− 𝑝𝑝
𝑀𝑀 𝑀𝑀 𝑀𝑀 𝑀𝑀

2
 Decision Boundary and P(C) for M-ary PAM
𝑑𝑑
• p = filled area = tail end of Gaussian density = 𝑄𝑄
2𝑁𝑁𝑜𝑜
•d=1
0.4

0.3

0.2

0.1

0
-5 -4 -3 -2 -1 0 1 2 3 4 5

3
 Probability of Error for M-ary PAM
• Average energy per symbol is given by

2 2 2 2
2 𝑑𝑑 3𝑑𝑑 5𝑑𝑑 𝑀𝑀 − 1 𝑑𝑑
𝐸𝐸𝑠𝑠 = + + + ⋯+
𝑀𝑀 2 2 2 2
𝑀𝑀
2
2 𝑑𝑑2 2 2 2
𝑑𝑑 2
= 1+ 3 + 5 + ⋯ + 𝑀𝑀 − 1 = � 4𝑚𝑚2 − 4𝑚𝑚 + 1
𝑀𝑀 4 2𝑀𝑀
𝑚𝑚=1
𝑑𝑑 2
= 𝑀𝑀2 − 1
12
𝑛𝑛 𝑛𝑛(𝑛𝑛+1) 𝑛𝑛 2 𝑛𝑛(𝑛𝑛+1)(2𝑛𝑛+1)
• Notice that 𝑘𝑘=1 𝑘𝑘 =
∑ ∑
, 𝑘𝑘=1 𝑘𝑘 =
2 6

4
 Probability of Error for M-ary PAM
𝑑𝑑 2 𝑀𝑀
• The total energy of PAM is 𝐸𝐸𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 = 𝑀𝑀2 − 1
12
𝑑𝑑 2 12𝐸𝐸𝑠𝑠
• Solving 𝐸𝐸𝑠𝑠 = 𝑀𝑀2 − 1 for d, we have 𝑑𝑑 = 2 .
12 𝑀𝑀 −1
𝑑𝑑
• Substituting d into p, we get (𝑝𝑝 = 𝑄𝑄 )
2𝑁𝑁𝑜𝑜
12𝐸𝐸𝑠𝑠
1 1 𝑑𝑑 1 𝑀𝑀2 −1
• 𝑃𝑃 𝐸𝐸 = 2 1 − 𝑝𝑝 = 2 1 − 𝑄𝑄 =2 1− 𝑄𝑄
𝑀𝑀 𝑀𝑀 2𝑁𝑁𝑜𝑜 𝑀𝑀 2𝑁𝑁𝑜𝑜
1 6𝐸𝐸𝑠𝑠 1 6 𝑙𝑙𝑙𝑙𝑙𝑙2 𝑀𝑀 𝐸𝐸𝑏𝑏
=2 1− 𝑄𝑄 =2 1− 𝑄𝑄
𝑀𝑀 𝑀𝑀2 − 1 𝑁𝑁𝑜𝑜 𝑀𝑀 𝑀𝑀2 − 1 𝑁𝑁𝑜𝑜

5
 EXAMPLE
• Find the probability of error for 8 PAM (M = 8). Also find the numerical
value of the probability of error when Eb/No = 15 dB.
format long e;
M=8
EbNodB=15
EbNo=10^(EbNodB/10)
Pe=(2*(1-1/M))*qfunc(sqrt((6*log2(M)*EbNo/(M^2-1))))

• Pe =
2.317416541261364e-003
6
 Probability of Error for Square QAM
• M = 16, 64, 256, 1024
• The constellation has 𝑀𝑀 rows and 𝑀𝑀 columns.
• The vectors of each row consists of
-( 𝑀𝑀 -1)d/2 … -3d/2 -d/2 d/2 3d/2 … ( 𝑀𝑀 -1)d/2
• The probability of correct is given by (inside, 4 sides, 4 corners)
2
𝑀𝑀−2 2 4 𝑀𝑀−2 4
• 𝑃𝑃 𝐶𝐶 = (1 − 2𝑝𝑝) + 1 − 𝑝𝑝 1 − 2𝑝𝑝 + (1 − 𝑝𝑝)2
𝑀𝑀 𝑀𝑀 𝑀𝑀
1 1 𝑑𝑑
• Ignoring p2 term, 𝑃𝑃 𝐸𝐸 = 1 − 𝑃𝑃 𝐶𝐶 = 4 1 − 𝑀𝑀
𝑝𝑝 = 4 1 −
𝑀𝑀
𝑄𝑄
2𝑁𝑁𝑜𝑜

7
16QAM
Q

8
 Probability of Error for Square QAM
• From the result for M-ary PAM, the energy for each row in the I direction is
𝑑𝑑 2 𝑀𝑀
given by (M→ 𝑀𝑀) 𝐸𝐸𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡,𝑜𝑜𝑜𝑜𝑜𝑜 𝑟𝑟𝑟𝑟𝑟𝑟 = 𝑀𝑀 − 1 .
12
𝑑𝑑 2 𝑀𝑀
• Since there are 𝑀𝑀 rows, 𝐸𝐸𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡,𝑎𝑎𝑎𝑎𝑎𝑎 𝑟𝑟𝑟𝑟𝑟𝑟𝑠𝑠 = 𝑀𝑀 − 1
12
• Since the energy of all columns is equal to the energy of all columns,
𝑑𝑑 2 𝑀𝑀
𝐸𝐸𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡,𝑎𝑎𝑎𝑎𝑎𝑎 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 𝑀𝑀 − 1
6
𝑑𝑑 2
• The average energy per symbol is given by 𝐸𝐸𝑠𝑠 = 𝑀𝑀 − 1
6
6𝐸𝐸𝑠𝑠
• 𝑑𝑑 =
𝑀𝑀−1

9
 Probability of Error for Square QAM
6𝐸𝐸𝑠𝑠
1 𝑑𝑑 1 𝑀𝑀−1
• 𝑃𝑃 𝐸𝐸 = 4 1 − 𝑄𝑄 =4 1− 𝑄𝑄
𝑀𝑀 2𝑁𝑁𝑜𝑜 𝑀𝑀 2𝑁𝑁𝑜𝑜
1 3𝐸𝐸𝑠𝑠 1 3𝑙𝑙𝑙𝑙𝑙𝑙2 (𝑀𝑀) 𝐸𝐸𝑏𝑏
• 𝑃𝑃 𝐸𝐸 = 4 1 − 𝑄𝑄 =4 1− 𝑄𝑄
𝑀𝑀 (𝑀𝑀−1)𝑁𝑁𝑜𝑜 𝑀𝑀 (𝑀𝑀−1) 𝑁𝑁𝑜𝑜

10
Probability of Error of 16QAM, 64QAM, and
256QAM
Prob of Error of 16QAM, 64QAM, 256QAM
0

-1

-2

-3

(power of 10) -4

-5

-6
e
P

-7

-8
16QAM
64QAM
-9
256QAM

-10
0 5 10 15 20 25
Eb/No (dB)

11
 EXAMPLE
• Find the probability of error for 256 QAM when Eb/No = 25 dB.
format long e;
M=256
EbNodB=25
EbNo=10^(EbNodB/10)
Pe=(4*(1-1/sqrt(M)))*qfunc(sqrt((3*log2(M)*EbNo/(M-1))))

• Pe =
9.155977668350282e-008
12
 Probability of Error for M-ary PSK
𝜋𝜋
• Minimum distance between the vectors = dmin = 2 𝐸𝐸𝑠𝑠 𝑠𝑠𝑠𝑠𝑠𝑠
𝑀𝑀
• The probability of correct when s1 is transmitted:
𝑃𝑃 𝐶𝐶 𝒔𝒔1
2
𝑥𝑥 − 𝐸𝐸𝑠𝑠 𝑦𝑦 2
𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑁𝑁 𝜋𝜋
𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑁𝑁
𝑜𝑜 𝑜𝑜
∞ 2 𝑥𝑥 𝑡𝑡𝑡𝑡𝑡𝑡 𝑀𝑀 2
= 2� 2 � 2 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
0 𝑁𝑁𝑜𝑜 0 𝑁𝑁𝑜𝑜
2𝜋𝜋 2𝜋𝜋
2 2
• P(E) = 1 - 𝑃𝑃 𝐶𝐶 𝒔𝒔1

13
Probability of Error for M-ary PSK
Q

𝐸𝐸𝑠𝑠

s1
I

14
 Probability of Error for M-ary PSK
• The probability of error P(E) is greater than the probability of error
between the two neighboring signal vectors. The probability of error
between the two neighboring signal vectors (assuming only two
neighboring vectors exist) with distance dmin is
𝑑𝑑𝑚𝑚𝑚𝑚𝑚𝑚
• 𝑃𝑃 𝐸𝐸 ≥ 𝑄𝑄
2𝑁𝑁𝑜𝑜
• As can be seen from the graph shown below, the probability of error P(E) is
𝑑𝑑𝑚𝑚𝑚𝑚𝑚𝑚
smaller than two times 𝑄𝑄 due to overlap.
2𝑁𝑁𝑜𝑜
𝑑𝑑𝑚𝑚𝑚𝑚𝑚𝑚 𝑑𝑑𝑚𝑚𝑚𝑚𝑚𝑚
• Thus, the probability of error is bounded: 𝑄𝑄 ≤ 𝑃𝑃(𝐸𝐸) ≤ 2𝑄𝑄
2𝑁𝑁𝑜𝑜 2𝑁𝑁𝑜𝑜

15
Probability of Error for M-ary PSK
Q

𝐸𝐸𝑠𝑠

s2

s1
I

sM

16
 BER for M-ary PSK (Alternate Solution)
• Find sk with phase closest to the phase of r.
• Let the transmitted signal be 𝒔𝒔1 = 𝐸𝐸𝑠𝑠 and the received signal be r.
• The received vector r can be represented as
• 𝒓𝒓 = 𝑟𝑟1 + 𝑗𝑗𝑟𝑟2 = 𝑅𝑅𝑒𝑒 𝑗𝑗𝜃𝜃𝑟𝑟
• 𝑅𝑅 = 𝑟𝑟12 + 𝑟𝑟22 r2 R

−1 𝑟𝑟2 θr s1
• 𝜃𝜃𝑟𝑟 = 𝑡𝑡𝑡𝑡𝑡𝑡
𝑟𝑟1 r1
𝐸𝐸𝑠𝑠
• 𝑟𝑟1 = 𝑅𝑅𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝑟𝑟 ), 𝑟𝑟2 = 𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖(𝜃𝜃𝑟𝑟 )
• 𝑟𝑟1 = 𝐸𝐸𝑠𝑠 + 𝑛𝑛1 , 𝑟𝑟1 = 𝑛𝑛2
17
 BER for M-ary PSK
• n1 and n2 are statistically independent random variables with
Gaussian pdfs. The mean value is zero and the variance is No/2.
• r1 is a Gaussian random variable with mean 𝐸𝐸𝑠𝑠 and variance No/2. r2
is a Gaussian random variable with mean zero and variance No/2.
• The joint pdf of r1 and r2 is given by
2
1 𝑟𝑟1 − 𝐸𝐸𝑠𝑠 +𝑟𝑟22 1 𝑟𝑟12 +𝑟𝑟22 −2 𝐸𝐸𝑠𝑠 𝑟𝑟1 +𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝑟𝑟1 , 𝑟𝑟2 = 𝑁𝑁 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑁𝑁 = 𝑁𝑁 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑁𝑁
2𝜋𝜋 2𝑜𝑜 2 2𝑜𝑜 2𝜋𝜋 2𝑜𝑜 2 2𝑜𝑜

18
 BER for M-ary PSK
• Changing the variable from r1, r2 (Cartesian coordinates) to R, θr (polar
coordinates), we have
𝑅𝑅 𝑅𝑅 2 −2 𝐸𝐸𝑠𝑠 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 )+𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝑅𝑅, 𝜃𝜃𝑟𝑟 = 𝑁𝑁 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑁𝑁
2𝜋𝜋 𝑜𝑜 2 2𝑜𝑜
2
• 𝑑𝑑𝑟𝑟1 𝑑𝑑𝑟𝑟2 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝜃𝜃𝑟𝑟 , R is Jacobian. 𝑟𝑟1 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 ), 𝑟𝑟2 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 )
𝜕𝜕𝑟𝑟1 𝜕𝜕𝑟𝑟2 𝜕𝜕𝑟𝑟1 𝜕𝜕𝑟𝑟2
• = 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝑟𝑟 ), = 𝑠𝑠𝑠𝑠𝑠𝑠(𝜃𝜃𝑟𝑟 ), = −𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 ), = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 )
𝜕𝜕𝑅𝑅 𝜕𝜕𝜕𝜕 𝜕𝜕𝜃𝜃𝑟𝑟 𝜕𝜕𝜃𝜃𝑟𝑟
𝜕𝜕𝑟𝑟1 𝜕𝜕𝑟𝑟1
𝜕𝜕𝜕𝜕 𝜕𝜕𝜃𝜃𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝑟𝑟 ) −𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 )
• 𝐽𝐽 = = = 𝑅𝑅
𝜕𝜕𝑟𝑟2 𝜕𝜕𝑟𝑟2 𝑐𝑐𝑐𝑐𝑐𝑐(𝜃𝜃𝑟𝑟 ) 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 )
𝜕𝜕𝜕𝜕 𝜕𝜕𝜃𝜃𝑟𝑟

19
 BER for M-ary PSK
• Integrating 𝑝𝑝 𝑅𝑅, 𝜃𝜃𝑟𝑟 for R from zero to infinity results in the marginal
density function
∞ 𝑅𝑅 𝑅𝑅 2 −2 𝐸𝐸𝑠𝑠 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅(𝜃𝜃𝑟𝑟 )+𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = ∫0 𝑁𝑁𝑜𝑜 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑁𝑁 𝑑𝑑𝑑𝑑
2𝜋𝜋 2
2 2𝑜𝑜

′ 𝑅𝑅 ′2 𝑅𝑅 2 𝑁𝑁𝑜𝑜
• Let 𝑅𝑅 = . Then, 𝑅𝑅 = 𝑁𝑁𝑜𝑜 , 𝑑𝑑𝑑𝑑 = 𝑑𝑑𝑅𝑅′
𝑁𝑁𝑜𝑜 2
2
2
∞ 𝑅𝑅′ 1 2𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = ∫0 2𝜋𝜋 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑅𝑅′2 −2 𝑅𝑅𝑅𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 + 𝑑𝑑𝑑𝑑′
2 𝑁𝑁𝑜𝑜 𝑁𝑁𝑜𝑜

20
 BER for M-ary PSK
• Changing 𝑅𝑅′ back to R and multiplying cos2(θr) + sin2(θr) = 1 to the
last term in the exponent, we have
∞ 𝑅𝑅 1 2𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = ∫0 2𝜋𝜋 𝑒𝑒𝑒𝑒𝑒𝑒 �− 2 �𝑅𝑅2 −2
𝑁𝑁𝑜𝑜
𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝜃𝜃𝑟𝑟 +
𝑁𝑁𝑜𝑜
𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜃𝜃𝑟𝑟 +
2𝐸𝐸𝑠𝑠
𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 �� 𝑑𝑑𝑑𝑑
𝑁𝑁𝑜𝑜

• Completing the square in the exponent, we get

2
∞ 𝑅𝑅 1 2𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = ∫0 2𝜋𝜋 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑅𝑅 − 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 + 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 𝑑𝑑𝑑𝑑
2 𝑁𝑁𝑜𝑜 𝑁𝑁𝑜𝑜

21
 BER for M-ary PSK
• Factoring out the second term in the exponent, we obtain
2
𝐸𝐸𝑠𝑠 ∞ 𝑅𝑅 1 2𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 ∫0 2𝜋𝜋 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑅𝑅 − 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 𝑑𝑑𝑑𝑑
𝑁𝑁𝑜𝑜 2 𝑁𝑁𝑜𝑜

• The probability of correct decision, which is the same as the

probability of correct decision when s1 is sent, is obtained by
integrating 𝑝𝑝 𝜃𝜃𝑟𝑟 from -π/M to π/M.
𝜋𝜋/𝑀𝑀
• 𝑃𝑃 𝐸𝐸 = 1 − 𝑃𝑃 𝐶𝐶 = 1 − ∫−𝜋𝜋/𝑀𝑀 𝑝𝑝 𝜃𝜃𝑟𝑟 𝑑𝑑𝜃𝜃𝑟𝑟
• P(E) can be evaluated using numerical integration.

22
 BER for M-ary PSK
2𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
• Let 𝑢𝑢 = 𝑅𝑅 − 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 . Then, 𝑅𝑅 = 𝑢𝑢 + 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 . 𝑑𝑑𝑑𝑑 = 𝑑𝑑𝑑𝑑.
𝑁𝑁𝑜𝑜 𝑁𝑁𝑜𝑜
2
∞ 𝑅𝑅 1 2𝐸𝐸𝑠𝑠
• ∫0 2𝜋𝜋 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑅𝑅 − 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 𝑑𝑑𝑑𝑑
2 𝑁𝑁𝑜𝑜

∞ 1 2𝐸𝐸𝑠𝑠 1 2
•= ∫ 2𝐸𝐸𝑠𝑠 𝑢𝑢 + 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑢𝑢 𝑑𝑑𝑢𝑢
− 𝑁𝑁 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 2𝜋𝜋 𝑁𝑁𝑜𝑜 2
𝑜𝑜
∞ 1 1
•= ∫ 2𝐸𝐸𝑠𝑠 𝑢𝑢 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑢𝑢2 𝑑𝑑𝑑𝑑 +
− 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 2𝜋𝜋
𝑁𝑁𝑜𝑜
2

𝐸𝐸𝑠𝑠 ∞ 1 1 2
𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 ∫ 2𝐸𝐸𝑠𝑠 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑢𝑢 𝑑𝑑𝑑𝑑
𝜋𝜋𝑁𝑁𝑜𝑜 − 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 2𝜋𝜋 2
𝑁𝑁𝑜𝑜

23
 BER for M-ary PSK
1 1 ∞
•= (−1) 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑢𝑢2 � 2𝐸𝐸
2𝜋𝜋 2 − 𝑁𝑁 𝑠𝑠 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟
𝑜𝑜

𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
•+ 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 1 − 𝑄𝑄 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟
𝜋𝜋𝜋𝜋𝑜𝑜 𝑁𝑁𝑜𝑜

1 𝐸𝐸𝑠𝑠 𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠

•= 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜃𝜃𝑟𝑟 + 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 1 − 𝑄𝑄 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟
2𝜋𝜋 𝑁𝑁𝑜𝑜 𝜋𝜋𝜋𝜋𝑜𝑜 𝑁𝑁𝑜𝑜

𝐸𝐸𝑠𝑠 1 𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑐𝑐𝑐𝑐𝑐𝑐 2 𝜃𝜃𝑟𝑟
𝑁𝑁𝑜𝑜 2𝜋𝜋 𝑁𝑁𝑜𝑜

𝐸𝐸𝑠𝑠 𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠

• +𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 1 − 𝑄𝑄 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟
𝑁𝑁𝑜𝑜 𝜋𝜋𝜋𝜋𝑜𝑜 𝑁𝑁𝑜𝑜

24
 BER for M-ary PSK
1 𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = 𝑒𝑒𝑒𝑒𝑒𝑒 −
2𝜋𝜋 𝑁𝑁𝑜𝑜
𝐸𝐸𝑠𝑠 𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
• +𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 1 − 𝑄𝑄 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟
𝑁𝑁𝑜𝑜 𝜋𝜋𝜋𝜋𝑜𝑜 𝑁𝑁𝑜𝑜

𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠 𝐸𝐸𝑠𝑠

• If ≫ 1, 𝑄𝑄 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 ≈ 0, 𝑒𝑒𝑒𝑒𝑒𝑒 − ≈ 0.
𝑁𝑁𝑜𝑜 𝑁𝑁𝑜𝑜 𝑁𝑁𝑜𝑜

2𝐸𝐸𝑠𝑠 1 𝐸𝐸𝑠𝑠
• 𝑝𝑝 𝜃𝜃𝑟𝑟 = 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟
𝑁𝑁𝑜𝑜 2𝜋𝜋 𝑁𝑁𝑜𝑜

25
 BER for M-ary PSK
𝜋𝜋/𝑀𝑀
• From 𝑃𝑃 𝐸𝐸 = 1 − 𝑃𝑃 𝐶𝐶 = 1 − ∫−𝜋𝜋/𝑀𝑀 𝑝𝑝 𝜃𝜃𝑟𝑟 𝑑𝑑𝜃𝜃𝑟𝑟 ,
𝜋𝜋 2𝐸𝐸𝑠𝑠 1 𝐸𝐸𝑠𝑠
• 𝑃𝑃 𝐸𝐸 ≈ 2 ∫𝜋𝜋/𝑀𝑀 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑠𝑠𝑠𝑠𝑠𝑠2 𝜃𝜃𝑟𝑟 𝑑𝑑𝜃𝜃𝑟𝑟
𝑁𝑁𝑜𝑜 2𝜋𝜋 𝑁𝑁𝑜𝑜

2𝐸𝐸𝑠𝑠 2𝐸𝐸𝑠𝑠
• Let 𝑣𝑣 = 𝑠𝑠𝑠𝑠𝑠𝑠 𝜃𝜃𝑟𝑟 . 𝑑𝑑𝑣𝑣 = 𝑐𝑐𝑐𝑐𝑐𝑐 𝜃𝜃𝑟𝑟 𝑑𝑑𝜃𝜃𝑟𝑟
𝑁𝑁𝑜𝑜 𝑁𝑁𝑜𝑜

𝜋𝜋 1 𝑣𝑣 2 2𝐸𝐸𝑠𝑠 𝜋𝜋
• 𝑃𝑃 𝐸𝐸 ≈ 2 ∫ 2𝐸𝐸𝑠𝑠 𝜋𝜋 𝑒𝑒𝑒𝑒𝑒𝑒 − 𝑑𝑑𝑣𝑣 = 2𝑄𝑄 𝑠𝑠𝑠𝑠𝑠𝑠
𝑠𝑠𝑠𝑠𝑠𝑠 2𝜋𝜋 2 𝑁𝑁𝑜𝑜 𝑀𝑀
𝑁𝑁𝑜𝑜 𝑀𝑀

• This is the upper bound of the probability of error for M-ary PSK.

26