Solution: The average symbol energy γs = 4 · 101.5 = 126.49.

The exact Ps is then given by

r !!2
2(4 − 1) 3.126.49
Ps = 1 − 1− Q = 7.37 · 10−7 .
4 15
The nearest neighbor approximation is given by
r !
3 · 126.49
Ps ≈ 4Q = 9.82 · 10−7 ,
15
which is slightly larger than the exact value owing to the conservative approximation that every constellation point
has four nearest neighbors. The symbol error probability for 16-PSK from Example 6.2 is Ps ≈ 1.916 · 10−3 ,
which is roughly four orders of magnitude larger than the exact Ps for 16-QAM. The larger Ps for MPSK versus
MQAM with the same M and same γb is due to the fact that MQAM uses both amplitude and phase to encode data
whereas MPSK uses just the phase. Thus, for the same energy per symbol or bit, MQAM makes more efficient use
of energy and therefore has better performance.

The MQAM demodulator requires both amplitude and phase estimates of the channel so that the decision regions
used in detection to estimate the transmitted symbol are not skewed in amplitude or phase. The analysis of perfor-
mance degradation due to phase estimation error is similar to the case of MPSK discussed previously. The channel
amplitude is used to scale the decision regions so that they correspond to the transmitted symbol: this scaling is
called automatic gain control (AGC). If the channel gain is estimated in error then the AGC improperly scales
the received signal, which can lead to incorrect demodulation even in the absence of noise. The channel gain is
typically obtained using pilot symbols to estimate the channel gain at the receiver. However, pilot symbols do not
lead to perfect channel estimates, and the estimation error can lead to bit errors. More details on the impact of
amplitude and phase estimation errors on the performance of MQAM modulation can be found in [7, Chap. 10.3;
8].

6.1.5 Error Probability for FSK and CPFSK

Let us first consider the error probability of binary FSK with the coherent demodulator of Figure 5.24. Since
demodulation is coherent, we can neglect any phase offset in the carrier signals. The transmitted signal is defined
by
√
si (t) = A 2Tb cos(2πfit), i = 1, 2. (6.26)
Hence Eb = A2 and γb = A2 /N0 . The input to the decision device is

z = s1 + n1 − s2 − n2 . (6.27)
The device outputs a 1-bit if z > 0 or a 0-bit if z ≤ 0. Let us assume that s1 (t) is transmitted; then

z|1 = A + n1 − n2 . (6.28)
An error occurs if z = A + n1 − n2 ≤ 0. On the other hand, if s2 (t) is transmitted then

z|0 = n1 − A − n2 , (6.29)

191
and an error occurs if z = n1 − A − n2 > 0. For n1 and n2 independent white Gaussian random variables with
mean zero and variance N0 /2, their difference is a white Gaussian random variable with mean zero and variance
equal to the sum of variances N0 /2 + N0 /2 = N0 . Then, for equally likely bit transmissions,
p √
Pb = .5p(A + n1 − n2 ≤ 0) + .5p(n1 − A − n2 > 0) = Q(A/ N0 ) = Q( γb ). (6.30)
The derivation of Ps for coherent MFSK with M > 2 is more complex and does not lead to a closed-form solution
[2, eq. (4.92)]. The probability of symbol error for noncoherent MFSK is derived in [9, Chap. 8.1] as
M −1    
X
m+1 M −1 1 −mγs
Ps = (−1) exp . (6.31)
m m+1 m+1
m=1

The error probability of CPFSK depends on whether the detector is coherent or noncoherent and also on
whether it uses symbol-by-symbol detection or sequence estimation. Analysis of error probability for CPFSK is
complex because the memory in the modulation requires error probability analysis over multiple symbols. The
formulas for error probability can also become quite complicated. Detailed derivations of error probability for
these different CPFSK structures can be found in [1; Chap. 5.3]. As with linear modulations, FSK performance
degrades under frequency and timing errors. A detailed analysis of the impact of such errors on FSK performance
can be found in [2, Chap. 5.2; 10; 11].

Table 6.1: Approximate symbol and bit error probabilities for coherent modulations
Modulation Ps (γs) Pb (γb)
√
BFSK Pb = Q( γ b )
√
BPSK Pb = Q( 2γb)
√ √
QPSK, 4-QAM Ps ≈ 2Q( γs ) ! Pb ≈ Q( 2γb) !
r r
2(M − 1) 6γ s 2(M − 1) 6γ b log2 M
MPAM Ps = Q Pb ≈ Q
M M2 − 1 M log2 M M2 − 1
p  π  2 p  π 
MPSK Ps ≈ 2Q 2γs sin Pb ≈ Q 2γb log2 M sin
M log2 M r ! M
q 
3γs 4 3γ b log2 M
Rectangular MQAM Ps ≈ 4Q M −1 Pb ≈ Q
log2 M M −1
q  r !
3γs 4 3γ b log2 M
Nonrectangular MQAM Ps ≈ 4Q M −1 Pb ≈ Q
log2 M M −1

6.1.6 Error Probability Approximation for Coherent Modulations

Many of the approximations or exact values for Ps derived so far for coherent modulation are in the following
form:
p
Ps (γs ) ≈ αM Q( βM γs ), (6.32)
where αM and βM depend on the type of approximation and the modulation type. In particular, the nearest
neighbor approximation has this form, where αM is the number of nearest neighbors to a constellation at the
minimum distance and βM is a constant that relates minimum distance to average symbol energy. In Table 6.1 we
summarize the specific values of αM and βM for common Ps expressions for PSK, QAM, and FSK modulations
based on the derivations in prior sections.

192