![](/user_photo/2706_HbeT2.jpg)
Fundamentals Of Wireless Communication
.pdf![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N401x1.jpg)
378 |
MIMO II: capacity and multiplexing architectures |
transmit antenna, with the other transmit antennas silent. At high SNR, the k pilot symbols allow the receiver to partially estimate the channel: over the nth block, k of the nt columns of Hn are estimated with a high degree of accuracy. This allows us to reliably communicate on the k × nr MIMO channel with receiver CSI.
1.Argue that the rate of reliable communication using this scheme at high SNR is approximately at least
|
Tc − k |
|
mink n |
log SNR bits/s/Hz |
(8.109) |
|
|||||
Tc |
r |
|
|
Hint: An information theory fact says that replacing the effect of channel uncertainty as Gaussian noise (with the same covariance) can only make the reliable communication rate smaller.
2.Show that the optimal training time (and the corresponding number of transmit antennas to use) is
k = min nt nr |
T |
|
|
2c |
(8.110) |
Substituting this in (8.109) we see that the number of spatial degrees of freedom using the pilot scheme is equal to
|
Tc |
|
|
|
Tc − k |
k |
(8.111) |
|
|
3.A reading exercise is to study [155], which shows that the capacity of the noncoherent block fading channel at high SNR also has the same number of spatial degrees freedom as in (8.111).
Exercise 8.9 Consider the time-invariant frequency-selective MIMO channel:
|
L−1 |
|
ym = |
|
|
H xm − + wm |
(8.112) |
=0
Construct an appropriate OFDM transmission and reception scheme to transform the original channel to the following parallel MIMO channel:
˜ |
= |
˜ |
+ ˜ |
n |
= |
0 Nc |
− |
1 |
(8.113) |
yn |
|
H˜ nxn |
wn |
|
|
N ˜ n = N −
Here c is the number of OFDM tones. Identify Hn, 0 c 1 in terms of
H = 0 L − 1.
Exercise 8.10 Consider a fixed physical environment and a corresponding flat fading MIMO channel. Now suppose we double the transmit power constraint and the bandwidth. Argue that the capacity of the MIMO channel with receiver CSI exactly doubles. This scaling is consistent with that in the single antenna AWGN channel.
Exercise 8.11 Consider (8.42) where independent data streams xim are transmitted on the transmit antennas (i = 1 nt ):
|
nt |
|
ym = |
|
|
hixim + wm |
(8.114) |
i=1
Assume nt ≤ nr .
![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N402x1.jpg)
379 |
8.7 Exercises |
1.We would like to study the operation of the decorrelator in some detail here. So
we make the assumption that hi is not a linear combination of the other vectors
h1 hi−1 hi+1 hnt for every i = 1 nt . Denoting H = h1 · · · hnt , show that this assumption is equivalent to the fact that H H is invertible.
2.Consider the following operation on the received vector in (8.114):
xˆm = H H −1H ym |
(8.115) |
= xm + H H −1H wm |
(8.116) |
Thus xˆim = xim + w˜ im where w˜ m = H H −1H wm is colored Gaussian noise. This means that the ith data stream sees no interference from any of the other streams in the received signal xˆim. Show that xˆim must be the output of the decorrelator (up to a scaling constant) for the ith data stream and hence conclude the validity of (8.47). This property, and many more, about the decorrelator can be learnt from Chapter 5 of [99]. The special case of nt = nr = 2 can be verified by explicit calculations.
Exercise 8.12 Suppose H (with nt < nr ) has i.i.d. 0 1 entries and denote h1 hnt as the columns of H. Show that the probability that the columns are linearly dependent is zero. Hence, conclude that the probability that the rank of H is strictly smaller than nt is zero.
Exercise 8.13 Suppose H (with nt < nr ) has i.i.d. 0 1 entries and denote the columns of H as h1 hnt . Use the result of Exercise 8.12 to show that the dimension
of the subspace spanned by the vectors h1 hk−1 hk+1 hnt is nt − 1 with probability 1. Hence conclude that the dimension of the subspace Vk, orthogonal to
this one, has dimension nr − nt + 1 with probability 1.
Exercise 8.14 Consider the Rayleigh fading n × n MIMO channel H with i.i.d.
0 1 entries. In the text we have discussed a random matrix result about the
√
convergence of the empirical distribution of the singular values of H/ n. It turns out
√
that the condition number of H/ n converges to a deterministic limiting distribution. This means that the random matrix H is well-conditioned. The corresponding limiting density is given by
4 |
e−2/x |
2 |
|
|
|
f x = |
|
|
|
(8.117) |
|
x3 |
|
A reading exercise is to study the derivation of this result proved in Theorem 7.2 of [32].
Exercise 8.15 Consider communicating over the time-invariant nt ×nr MIMO channel:
ym = Hxm + wm |
(8.118) |
The information bits are encoded using, say, a capacity-achieving Gaussian code such as an LDPC code. The encoded bits are then modulated into the transmit signal xm; typically the components of the transmit vector belong to a regular constellation such as QAM. The receiver, typically, operates in two stages. The first stage is demodulation: at each time, soft information (a posteriori probabilities of the bits that modulated the
![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N403x1.jpg)
![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N404x1.jpg)
381 |
8.7 Exercises |
Exercise 8.19 Consider detection on the generic vector channel with additive colored Gaussian noise (cf. (8.72)).
1. Show that the output of the linear MMSE receiver,
vmmsey |
(8.123) |
is a sufficient statistic to detect x from y. This is a generalization of the scalar sufficient statistic extracted from the vector detection problem in Appendix A (cf. (A.55)).
2.From the previous part, we know that the random variables y and x are independent conditioned on vmmsey. Use this to verify (8.73).
Exercise 8.20 We have seen in Figure 8.13 that, at low SNR, the bank of linear matched filter achieves capacity of the 8 by 8 i.i.d. Rayleigh fading channel, in the sense that the ratio of the total achievable rate to the capacity approaches 1. Show that this is true for general nt and nr .
Exercise 8.21 Consider the n by n i.i.d. flat Rayleigh fading channel. Show that the total achievable rate of the following receiver architectures scales linearly with n: (a) bank of linear decorrelators; (b) bank of matched filters; (c) bank of linear MMSE receivers. You can assume that independent information streams are coded and sent out of each of the transmit antennas and the power allocation across antennas is uniform. Hint: The calculation involving the linear MMSE receivers is tricky. You have to show that the linear MMSE receiver performance, asymptotically for large n, depends on the covariance matrix of the interference faced by each stream only through its empirical eigenvalue distribution, and then apply the large-n random matrix result used in Section 8.2.2. To show the first step, compute the mean and variance of the output SINR, conditional on the spatial signatures of the interfering streams. This calculation is done in [132, 123]
Exercise 8.22 Verify (8.71) by direct matrix manipulations.
Hint: You might find useful the following matrix inversion lemma (for invertible A),
A + xx −1 = A−1 − |
A−1xx A−1 |
(8.124) |
1 + x A−1x |
Exercise 8.23 Consider the outage probability of an i.i.d. Rayleigh MIMO channel (cf. (8.81)). Show that its decay rate in SNR (equal to P/N0) is no faster than nt · nr by justifying each of the following steps.
pout R ≥ |
log det Inr + SNR HH < R |
(8.125) |
||||
≥ |
SNR Tr HH < R |
(8.126) |
||||
≥ |
SNR h11 2 < R nt nr |
(8.127) |
||||
|
|
1 − e− |
R nt nr |
(8.128) |
||
|
|
SNR |
|
|||
= Rnt nr |
|
|||||
≈ |
|
|
(8.129) |
|||
SNRnt nr |
382 |
MIMO II: capacity and multiplexing architectures |
Exercise 8.24 Calculate the maximum diversity gains for each of the streams in the V-BLAST architecture using the MMSE–SIC receiver. Hint: At high SNR, interference seen by each stream is very high and the SINR of the linear MMSE receiver is very close to that of the decorrelator in this regime.
Exercise 8.25 Consider communicating over a 2 × 2 MIMO channel using the D-BLAST architecture with N = 1 and equal power allocation P1 = P2 = P for both the layers. In this exercise, we will derive some properties of the parallel channel (with L = 2 diversity branches) created by the MMSE–SIC operation. We denote the MIMO channel by H = h1 h2 and the projections
|
h1 2 = |
h1 h2 |
|
h2 |
h1 2 = h1 |
− h1 2 |
|
|
(8.130) |
||||||
|
h2 2 |
|
|
|
|||||||||||
Let us denote the induced parallel channel as |
|
|
|
|
|
|
|
|
|||||||
|
|
y = g x + w |
= 1 2 |
|
|
|
|
(8.131) |
|||||||
1. Show that |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
g1 |
2 |
= h1 2 |
2 |
|
|
h1 2 2 |
g2 |
2 |
= h2 |
2 |
|
|
|||
|
|
+ |
|
SNR h2 2 + 1 |
|
|
|
|
(8.132) |
where SNR = P/N0.
2.What is the marginal distribution of g1 2 at high SNR? Are g1 2 and g2 2 positively correlated or negatively correlated?
3.What is the maximum diversity gain offered by this parallel channel?
4.Now suppose g1 2 and g2 2 in the parallel channel in (8.131) are independent, while still having the same marginal distribution as before. What is the maximum
diversity gain offered by this parallel channel?
Exercise 8.26 Generalize the staggered stream structure (discussed in the context of a 2 × nr MIMO channel in Section 8.5) of the D-BLAST architecture to a MIMO channel with nt > 2 transmit antennas.
Exercise 8.27 Consider a block length N D-BLAST architecture on a MIMO channel with nt transmit antennas. Determine the rate loss due to the initialization phase as a function of N and nt .
![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N406x1.jpg)
![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N407x1.jpg)
384 MIMO III: diversity–multiplexing tradeoff and universal space-time codes
the distribution of the fading channel gains. The drawback of the approach is that the performance of the designed codes may be sensitive to the supposed fading distribution. This is problematic, since, as we mentioned in Chapter 2, accurate statistical modeling of wireless channels is difficult. The outage formulation, however, suggests a different approach. The operational interpretation of the outage performance is based on the existence of universal codes: codes that simultaneously achieve reliable communication over every MIMO channel that is not in outage. Such codes are robust from an engineering point of view: they achieve the best possible outage performance for every fading distribution. This result motivates a universal code design criterion: instead of using the pairwise error probability averaged over the fading distribution of the channel, we consider the worst-case pairwise error probability over all channels that are not in outage. Somewhat surprisingly, the universal code-design criterion is closely related to the product distance, which is obtained by averaging over the Rayleigh distribution. Thus, the product distance criterion, while seemingly tailored for the Rayleigh distribution, is actually more fundamental. Using universal code design ideas, we construct codes that achieve the optimal diversity–multiplexing tradeoff.
Throughout this chapter, the receiver is assumed to have perfect knowledge of the channel matrix while the transmitter has none.
9.1 Diversity–multiplexing tradeoff
In this section, we use the outage formulation to characterize the performance capability of slow fading MIMO channels in terms of a tradeoff between diversity and multiplexing gains. This tradeoff is then used as a unified framework to compare the various space-time coding schemes described in this book.
9.1.1 Formulation
When we analyzed the performance of communication schemes in the slow fading scenario in Chapters 3 and 5, the emphasis was on the diversity gain. In this light, a key measure of the performance capability of a slow fading channel is the maximum diversity gain that can be extracted from it. For example, a slow i.i.d. Rayleigh faded MIMO channel with nt transmit and nr receive antennas has a maximum diversity gain of nt ·nr: i.e., for a fixed target rate R, the outage probability pout R decays like 1/SNRnt nr at high SNR.
On the other hand, we know from Chapter 7 that the key performance benefit of a fast fading MIMO channel is the spatial multiplexing capability it provides through the additional degrees of freedom. For example, the
385 |
9.1 Diversity–multiplexing tradeoff |
|
|
|
|
|
||
|
capacity of an i.i.d. Rayleigh fading channel scales like nmin log SNR, where |
|||||||
|
nmin = min nt nr is the number of spatial degrees of freedom in the chan- |
|||||||
|
nel. This fast fading (ergodic) capacity is achieved by averaging over the |
|||||||
|
variation of the channel over time. In the slow fading scenario, no such aver- |
|||||||
|
aging is possible and one cannot communicate at this rate reliably. Instead, |
|||||||
|
the information rate allowed through the channel is a random variable fluc- |
|||||||
|
tuating around the fast fading capacity. Nevertheless, one would still expect |
|||||||
|
to be able to benefit from the increased degrees of freedom even in the |
|||||||
|
slow fading scenario. Yet the maximum diversity gain provides no such |
|||||||
|
indication; for example, both an nt × nr |
channel and an ntnr |
× 1 channel |
|||||
|
have the same maximum diversity gain and yet one would expect the for- |
|||||||
|
mer to allow better spatial multiplexing than the latter. One needs something |
|||||||
|
more than the maximum diversity gain to capture the spatial multiplexing |
|||||||
|
benefit. |
|
|
|
|
|
|
|
|
|
Observe that to achieve the maximum diversity gain, one needs to com- |
||||||
|
municate at a fixed rate R, which becomes vanishingly small compared to |
|||||||
|
the fast fading capacity at high SNR (which grows like nmin log SNR). Thus, |
|||||||
|
one is actually sacrificing all the spatial multiplexing benefit of the MIMO |
|||||||
|
channel to maximize the reliability. To reclaim some of that benefit, one |
|||||||
|
would instead want to communicate at a rate R = r log SNR, which is a fraction |
|||||||
|
of the fast fading capacity. Thus, it makes sense to formulate the following |
|||||||
|
diversity–multiplexing tradeoff for a slow fading channel. |
|
||||||
|
|
|
|
|
||||
|
|
A diversity gain d r is achieved at multiplexing gain r if |
|
|||||
|
|
|
R = r log SNR |
|
|
(9.1) |
||
|
|
and |
|
|
|
|
|
|
|
|
|
pout R ≈ SNR−d r |
|
(9.2) |
|||
|
|
or more precisely, |
|
|
|
|
|
|
|
|
lim |
log pout r log SNR |
= − |
d |
(9.3) |
||
|
|
log SNR |
|
|
||||
|
|
SNR→ |
|
|
r |
|||
|
|
The curve d · is the diversity–multiplexing tradeoff of the slow fading |
||||||
|
|
channel. |
|
|
|
|
|
|
|
|
The above tradeoff characterizes the slow fading performance limit of the |
||||||
|
channel. Similarly, we can formulate a diversity–multiplexing tradeoff for |
|||||||
|
any space-time coding scheme, with outage probabilities replaced by error |
|||||||
|
probabilities. |
|
|
|
|
|
|
![](/html/2706/250/html_XnLboG3mqg.Gno5/htmlconvd-PNPQ3N409x1.jpg)
386 |
MIMO III: diversity–multiplexing tradeoff and universal space-time codes |
A space-time coding scheme is a family of codes, indexed by the signal- to-noise ratio SNR. It attains a multiplexing gain r and a diversity gain d if the data rate scales as
R = r log SNR |
|
(9.4) |
|||
and the error probability scales as |
|
|
|
||
pe ≈ SNR−d |
|
(9.5) |
|||
i.e., |
|
|
|
||
lim |
log pe |
= − |
d |
(9.6) |
|
|
|
||||
SNR→ log SNR |
|
|
The diversity–multiplexing tradeoff formulation may seem abstract at first sight. We will now go through a few examples to develop a more concrete feel. The tradeoff performance of specific coding schemes will be analyzed and we will see how they perform compared to each other and to the optimal diversity–multiplexing tradeoff of the channel. For concreteness, we use the i.i.d. Rayleigh fading model. In Section 9.2, we will describe a general approach to tradeoff-optimal space-time code based on universal coding ideas.
9.1.2 Scalar Rayleigh channel
PAM and QAM
Consider the scalar slow fading Rayleigh channel,
y m = hx m + w m |
(9.7) |
with the additive noise i.i.d. 0 1 and the power constraint equal to SNR. Suppose h is 0 1 and consider uncoded communication using PAM with a data rate of R bits/s/Hz. We have done the error probability analysis in Section 3.1.2 for R = 1; for general R, the analysis is similar. The average error probability is governed by the minimum distance between the PAM points. The constellation ranges from approximately −√SNR to +√SNR, and since there are 2R constellation points, the minimum distance is approximately
Dmin ≈ |
√SNR |
|
(9.8) |
2R |
387 9.1 Diversity–multiplexing tradeoff
and the error probability at high SNR is approximately (cf. (3.28)), |
|
|||||||||||||||||
|
|
|
|
− # |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
|
|
D2 |
1 |
|
|
|
22R |
|
|
||||||
pe |
|
|
1 |
|
min |
|
|
|
|
|
|
|
|
|
|
(9.9) |
||
≈ |
2 |
|
4 + Dmin2 |
≈ Dmin2 |
|
≈ SNR |
||||||||||||
|
|
|
|
|
|
|||||||||||||
By setting the data rate R = r log SNR, we get |
|
|
|
|
|
|
|
|||||||||||
|
|
|
|
|
1 |
|
|
|
|
|
|
|
|
|
|
|||
|
|
|
|
pe ≈ |
|
|
|
|
|
|
|
|
(9.10) |
|||||
|
|
|
|
SNR1−2r |
|
|
|
|
|
|
||||||||
yielding a diversity–multiplexing tradeoff of |
|
|
|
|
|
|
|
|||||||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
dpam r = 1 − 2r |
r 0 |
1 |
|
|
|
(9.11) |
|||||||||||
|
|
|
|
|
||||||||||||||
|
2 |
|
|
Note that in the approximate analysis of the error probability above, we focus on the scaling of the error probability with the SNR and the data rate but are somewhat careless with constant multipliers: they do not matter as far as the diversity–multiplexing tradeoff is concerned.
We can repeat the analysis for QAM with data rate R. There are now 2R/2 constellation points in each of the real and imaginary dimensions, and hence
the minimum distance is approximately |
|
|
|
|||||
|
|
|
√ |
|
|
|
|
|
|
Dmin ≈ |
SNR |
|
(9.12) |
||||
|
|
|
|
|||||
2R/2 |
|
|||||||
and the error probability at high SNR is approximately |
|
|||||||
|
pe ≈ |
|
2R |
|
|
(9.13) |
||
|
|
|
|
|
||||
SNR |
|
|
||||||
yielding a diversity–multiplexing tradeoff of |
|
|||||||
|
|
|
|
|
|
|
||
|
dqam r = 1 − r |
|
|
|
r 0 1 |
(9.14) |
The tradeoff curves are plotted in Figure 9.1.
Let us relate the two endpoints of a tradeoff curve to notions that we already know. The value dmax = d 0 can be interpreted as the SNR exponent that describes how fast the error probability can be decreased with the SNR for a fixed data rate; this is the classical diversity gain of a scheme. It is 1 for both PAM and QAM. The decrease in error probability is due to an increase in Dmin. This is illustrated in Figure 9.2.
In a dual way, the value rmax for which d rmax = 0 describes how fast the data rate can be increased with the SNR for a fixed error probability. This
number can be interpreted as the number of (complex) degrees of freedom that are exploited by the scheme. It is 1 for QAM but only 1/2 for PAM.