Haykin_ch09_pp3.

fm Page 569 Friday, January 4, 2013 4:58 PM

Problems 569

b. Show that the delay-spread function of the multipath channel is described by

h˜  ; t  = ̃  t    
where    is the Dirac delta function in the  -domain. Hence, justify the statement that the
channel described in this problem is a time-selective channel.
c. Let S  f  and S s̃  f  denote the Fourier transforms of ̃  t  and s̃  t  , respectively. What then is
̃
the Fourier transform of x̃  t  ?
d. Using the result of part c, justify the statement that the multipath channel described herein can be
approximately frequency-flat. What is the condition that would satisfy this description?
9.4 In this problem, we consider a multipath channel embodying large-scale effects. Specifically, using
complex notation, the received signal at the channel output is described by
L
x̃  t  =  ̃l s̃  t – l 
l=1
where ̃ l and  l denote the amplitude and time delay associated with the lth path in the channel for
l = 1, 2, , L. Note that ̃ l is assumed to be constant for all l.
a. Show that the delay-spread function of the channel is described by
L
h˜   ; t  =  ̃l    – l 
l=1
where    is the Dirac delta function expressed in the  -domain.
b. This channel is said to be time-nonselective. Why?
c. The channel does exhibit a frequency-dependent behavior. To illustrate this behavior, consider
the following delay-spread function:
h˜   ; t  =     + ̃ 2    –  2 
where  2 is the time delay produced by the second path in the channel. Plot the magnitude
(amplitude) response of the channel for the following specifications:
i. ̃ 2 = 0.5
ii. ̃ 2 = j  2
iii. ̃ 2 = – j
where j = – 1 . Comment on your results.
9.5 Expanding on the multipath channel considered in Problem 9.4, a more interesting case is
characterized by the scenario in which the received signal at the channel output is described as follows:
L
x̃  t  =  ̃l  t s̃  t – l  t  
l=1
where the amplitude ̃ l  t  and time delay  l  t  for the lth path are both time dependent for
l = 1, 2, , L.
a. Show that the delay-spread function of the multipath channel described herein is given by
L
h˜   ; t  =  ̃l  t     – l  t  
l=1
where    is the Dirac delta function in the  -domain. This channel is said to exhibit both
large- and small-scale effects. Why?
b. The channel is also said to be both time selective and frequency selective. Why?
Haykin_ch09_pp3.fm Page 570 Friday, January 4, 2013 4:58 PM

570 Chapter 9 Signaling over Fading Channels

c. To illustrate the point made under b, consider the following channel description:

h˜   ; t  = ̃ 1  t      + ̃ 2  t    t –  2 
where ̃ 1  t  and ̃ 2  t  are both Rayleigh processes.
For selected ̃ 1  t  , ̃ 2  t  and  2 , do the following:
i. At each time t = 0, compute the Fourier transform of h̃   ; t  .
ii. Hence, plot the magnitude spectrum of the channel, that is, H̃  f ;t  , expressed as a
function of both time t and frequency f.
Comment on the results so obtained.
9.6 Consider a multipath channel where the delay-spread function is described by
L
h̃   ; t  =  ̃l  t     – l 
l=1

where the scattering processes attributed to the time-varying amplitude ̃ l  t  and fixed delay  l are
uncorrelated for l = 1, 2, , L.
a. Determine the correlation function of the channel, namely R   1 t 1 ; 2 t 2  .
h̃
b. With a Jakes model for the scattering process described in (9.12), find the corresponding formula
for the correlation function of the channel under part a of the problem.
c. Hence, justify the statement that the multipath channel described in this problem fits a WSSUS
model.
9.7 Revisit the Jakes model for a fast fading channel described in (9.12). Let the coherence time be
defined as that range of values t over which the correlation function defined in (9.12) is greater
than 0.5.
For some prescribed maximum Doppler shift  max , find the coherence time of the channel.
9.8 Consider a multipath channel for which the delay-spread function is given by
L
h̃   ; t  =  ̃l  t    t – l 
l=1

where the amplitude ̃ l  t  is time varying but the time delay  l is fixed. As in Problem 9.4, the
scattering processes are described by the Jakes model in (9.12). Determine the power-delay profile
of the channel, P ˜   .
h
9.9 In real-life situations, the wireless channel is nonstationary due to the presence of moving objects of
different kinds and other physical elements that can significantly affect radio propagation. Naturally,
different types of wireless channels have different degrees of nonstationarity.
Even though many wireless communication channels are indeed highly nonstationary, the WSSUS
model described in Section 9.4 still provides a reasonably accurate account of the statistical
characteristics of the channel. Elaborate on this statement.

“Space Diversity-on-Receive” Systems

9.10 Following the material presented on Rayleigh fading in Chapter 4, derive the probability density
function of (9.64).
9.11 A receive-diversity system uses a selection combiner with two diversity paths. The outage occurs
when the instantaneous SNR  drops below 0.25av, where av is the average SNR.
Determine the probability of outage experienced by the receiver.
Haykin_ch09_pp3.fm Page 571 Friday, January 4, 2013 4:58 PM

Problems 571

9.12 The average SNR in a selection combiner is 20 dB. Compute the probability that the instantaneous
SNR of the selection combiner drops below  = 10 dB for the following number of receive antennas:
a. Nr = 1
b. Nr = 2
c. Nr = 3
d. Nr = 4.
Comment on your results.
9.13 Repeat Problem 9.12 for  = 15 dB.
9.14 In Section 9.8 we derived the optimum values of (9.75) for complex weighting factors of the
maximal-ratio combiner using the Cauchy–Schwartz inequality.
This problem addresses the same issue, but this time we use the standard maximization procedure.
To simplify matters, the number of diversity paths Nr is restricted to two, with the complex
weighting parameters denoted by a1 and a2. Let
a k = x k + jy k k = 1 2
The complex derivative with respect to ak is defined by
 1  
-------- = ---  -------- + j --------   k = 1 2
a *k 2  x k y k 

Applying this formula to the combiner’s output SNR c of (9.71), derive the optimum  mrc in (9.75).
9.15 As discussed in Section 9.8, an equal-gain combiner is a special form of the maximal-ratio combiner
for which the weighting factors are all equal. For convenience of presentation, the weighting
parameters are set to unity.
Assuming that the instantaneous SNR  is small compared with the average SNR  av , derive an
approximate formula for the probability density function of the random variable  represented by
the sample  .
9.16 Compare the performances of the following linear “diversity-on-receive” techniques:
a. Selection combiner.
b. Maximal-ratio combiner.
c. Equal-gain combiner.
Base the comparison on signal-to-noise improvement, expressed in decibels for the following
number of diversity branches: Nr = 2, 3, 4, 5, 6.
9.17 Show that the maximum-likelihood decision rule for the maximal-ratio combiner may be formulated
in the following two equivalent forms:
a. If

   12 +  22  s i 2 – y 1 s i – y 1 s i      12 +  22  s k 2 – y 1 s k – y 1 s k  ki
then choose symbol si over sk.
b. If, by the same token,
2 2 2 2 2 2 2 2
   1 +  2 – 1  s i + d  y 1 s i       1 +  2 – 1  s k + d  y 1 s k   ki
2
then choose symbol si over sk. Here, d (y1,si) denotes the squared Euclidean distance between
the signal points y1 and si.
9.18 It may be argued that, in a rather loose sense, transmit-diversity and receive-diversity antenna
configurations are the dual of each other, as illustrated in Figure P9.18.
a. Taking a general viewpoint, justify the mathematical basis for this duality.
Haykin_ch09_pp3.fm Page 572 Tuesday, January 8, 2013 11:21 AM

572 Chapter 9 Signaling over Fading Channels

b. However, we may cite the example of frequency-division diplexing (FDD) for which, in a strict
sense, we find that the duality depicted in Figure P9.18 is violated. How is it possible for the
violation to arise in this example?

Diversity 1 1
paths

2 2 Receive
antenna

…
…

…
Transmit
antenna
Diversity
Nr Nr paths
Multiple Multiple
receive transmit
Figure P9.18 antennas antennas

“Space Diversity-on-Transmit” Systems

9.19 Show that the two-by-two channel matrix in (9.88), defined in terms of the multiplicative fading
j 1 j 2
factors  1 e and  2 e , is a unitary matrix, as shown by
†
j 1 j 2 j 1 j 2
1 e 2 e 1 e 2 e
=  1 + 2  1 0
2 2
–j 2 –j 1 –j 2 –j 1 0 1
2 e –1 e 2 e –1 e

9.20 Derive the formula for the average probability of symbol error incurred by the Alamouti code.
9.21 Figure P9.22 shows the extension of orthogonal space–time codes to the Alamouti code, using two
antennas on both transmit and receive. The sequence of signal encoding and transmissions is
identical to that of the single-receiver case of Figure 9.18. Part a of the table below defines the
channels between the transmit and receive antennas. Part b of the table defines the outputs of the
receive antennas at times t and t + T , where T is the symbol duration.
a. Derive expressions for the received signals x̃ 1 x̃ 2 x̃ 3 , and x̃ 4, including the respective additive
noise components expressed in terms of the transmitted symbols.
b. Derive expressions for the line of combined outputs in terms of the received signals.
c. Derive the maximum-likelihood decision rule for the estimates s̃ 1 and s̃ 2 .

Receive antenna 1 Receive antenna 2

a. Transmit antenna 1 h1 h3
Transmit antenna 2 h2 h4

b. Time t x̃ 1 x̃ 3
Time t + T x̃ 2 x̃ 4

9.22 This problem explores a new interpretation of the Alamouti code. Let
1 2
s̃ i = s i + js i  i = 1 2
Haykin_ch09_pp3.fm Page 573 Tuesday, January 8, 2013 11:21 AM

Problems 573

Figure P9.22 s˜1, s˜2,

–s˜2 s˜1

Transmit Transmit
antenna 1 antenna 1

h1 h2 h3 h4

Receive Receive
antenna 1 antenna 2

˜ 1,
w Interference w̃3, Interference
Σ Σ
w̃2 and noise w̃2 and noise

h1 h3
Channel Linear Channel
h2 h4
estimator 1 combiner estimator 2

h1 h2 y˜1 y˜ 2 h3 h4

Maximum-likelihood decoder

s˜1 s˜2

1 2
where s i and s i are both real numbers. The complex entry s̃ i in the 2-by-2 Alamouti code is
represented by the 2-by-2 real orthogonal matrix
1 2
si si
 i = 1 2
2 1
–si si

Likewise, the complex-conjugated entry s̃ *i is represented by the 2-by-2 real orthogonal matrix
1 1
si –si
 i = 1 2
2 2
si si

a. Show that the 2-by-2 complex Alamouti code S is equivalent to the 4-by-4 real transmission
matrix

1 2 1 2

s1 s1 s2 s2
2 1 2 1
–s1 s1 –s2 s2
S4 =
1 2 1 2
–s2 s2 s1 –s1
2 1 2 1
–s2 –s2 s1 s1
Haykin_ch09_pp3.fm Page 574 Friday, January 4, 2013 4:58 PM

574 Chapter 9 Signaling over Fading Channels

b. Show that S4 is an orthogonal matrix.

c. What is the advantage of the complex code S over the real code S4?
9.23 For two transmit antennas and simple receive antenna, the Alamouti code is said to be the only
optimal space–time block. Using the log-det formula of (9.117), justify this statement.
9.24 Show that the channel capacity of the Alamouti code is equal to the sum of the channel capacities of
two SISO systems with each one of them operating at half the original bit rate.

MIMO Wireless Communications

9.25 Show that, at high SNRs, the capacity gain of a MIMO wireless communication system with the
channel state known to the receiver is N = min{Nt ,Nr} bits per second per hertz for every 3 dB
increase in SNR.
9.26 To calculate the outage probability of MIMO systems, we use the complementary cumulative distri-
bution function of the random channel matrix H rather than the cumulative probability function itself.
Explain this rationale for calculating the outage probability.
9.27 Equation (9.120) defines the formula for the channel capacity of diversity-on-receive channel.
In Section 9.8 we pointed out that the selection combiner is a special case of the maximal-ratio
combiner. Using (9.120), formulate an expression for the channel capacity of wireless diversity
using the selection combiner.
9.28 For the special case of a MIMO system having Nt = Nr = N, show that the ergodic capacity of the
system scales linearly, rather than logarithmically, with increasing SNR as N approaches infinity.
9.29 In this problem we continue with the solution to Problem 9.28, namely
 av
C   -------------  as N  
 log e2

where Nt = Nr = N and  av is the average eigenvalue of the matrix produced HH† = H† H . What is
the value of the constant?
a. Justify the asymptotic result given in (9.119); that is,
C
----  constant
N
b. What conclusion can you draw from this asymptotic result?
9.30 Suppose that an additive, temporally stationary, Gaussian interference v(t) corrupts the basic
complex channel model of (9.105). The interference v(t) has zero mean and correlation matrix Rv.
Evaluate the effect of the interference v(t) on the ergodic capacity of the MIMO link.
9.31 Consider a MIMO link for which the channel may be considered to be essentially “constant for k
users of the channel.”
a. Starting with the basic channel model of (9.105), formulate the input–output relationship of this
link with the input being described by the Nr-by-k matrix
S =  s 1 s 2  s k 
b. How is the log-det capacity formula of the link correspondingly modified?
9.32 In a MIMO channel, the ability to exploit space-division multiple-access techniques for spectrally
efficient wireless communications is determined by the rank of the complex channel matrix H. (The
rank of a matrix is defined by the number of independent columns in the matrix.) For a given (Nt, Nr)
antenna configuration, it is desirable that the rank of H equal the minimum one of Nt transmit and Nr
receive antennas, for it is only then that we are able to exploit the full potential of the MIMO antenna