EEE 437 Wireless Communication

Diversity Techniques

Dr. Md. Forkan Uddin

Professor
Department of EEE, BUET
 What is Diversity?

Fading significantly degraded the system performance

 If the probability that the
channel gain is below a critical
level is P, then with L independent
channels, the probability is pL
 Diversity exploits the random
nature of radio propagation by
finding independent (or at least
highly uncorrelated) signal paths
for communication. 2
 What is Diversity?
 One of the most powerful techniques to mitigate the effects of fading
is to use diversity-combining of independently fading signal paths.

 Diversity-combining uses the fact that independent signal paths have

a low probability of experiencing deep fades simultaneously.

 The idea behind diversity is to send the same data over independent fading
paths (known as diversity branches).

 If the probability that the channel gain is below a critical level is p, then with L
independent channels, the probability is pL

 These independent paths are combined in some way such that the fading of
the resultant signal is reduced.

 Diversity techniques mitigating the effect of multi-path fading –

Microdiversity

 Diversity methods mitigating the effects of shadowing from buildings

and objects - Macrodiversity. 3
 Types of Diversity
1. Spatial/Space/Antenna diversity:
 Multiple transmit or receive antennas (an antenna array) are used, where the
elements of the array are separated in distance
 The same data is sent over independent fading paths
 No increase in transmit signal power or bandwidth
 Antenna separation must be enough to provide independent fading

 Channels from each antenna (h13, h23) to a third antenna

 Channels are uncorrelated when d12 > 1.5 λ
 Channels correlated, fade together when d12 ≤ λ
 This correlation distance depends on the radio environment around the pair
of antennas 4
 Types of Diversity
2. Time diversity achieved by transmitting the same signal at L different
times, where the time difference is greater than the channel coherence
time (the inverse of the channel Doppler spread)
Time diversity is achieved by averaging the fading of the channel over time. Typically,
the channel coherence time is of the order of tens to hundreds of symbols, and
therefore the channel is highly correlated across consecutive symbols. To ensure that
the coded symbols are transmitted through independent or nearly independent fading
gains, interleaving of codewords is required.
 Time diversity does not require increased transmit power, but it does decrease
the data rate since data is repeated in the diversity time slots rather than
sending new data in these time slots.
 Clearly time diversity can’t be used for stationary applications, since the
channel coherence time is infinite and thus fading is highly correlated over time.
 Time diversity is expensive. However, it is justified for critical traffic.

5
 Types of Diversity
3. Frequency diversity: Achieved by transmitting the same narrowband
signal with L different carrier frequencies, where the carriers are separated
by the coherence bandwidth of the channel.

Multipath fading: different effects on different frequencies

– Different subcarriers are independent of each other
• Frequency diversity is expensive. However, it is justified for critical traffic.

6
Types of Diversity

7
 Types of Diversity
4. Polarization diversity: Independent channels are implemented using the
fact that vertically and horizontally polarized paths are independent.

5. Angle diversity: Multiple receiver antennas with different directivity are

used to receive the same information-bearing signal at different angles.

8
 Spatial vs Time/Frequency Diversity
Spatial diversity
 No additional bandwidth required
 Increase of average SNR is possible
 Additional array gain is possible

Time/Frequency diversity
 Time/frequency is sacrificed
 No array gain
 Averaged receive SNR remains as that for
AWGN channel
9
 Space Diversity/ Diversity Combining
Most combining techniques are linear:
Output of the combiner is just a weighted sum of the
different fading paths or diversity branches.
 Selection Combining (SC)
 Maximal Ratio Combining (MRC)
 Equal gain combining (EGC)

General architecture of a linear combiner

10
 Selection Combining (SC) (1)
Principle: Select the diversity branch with the highest instantaneous SNR
 Only one branch is used at a time, so co-phasing of multiple branches is not
required
 But the RX must simultaneously and continuously monitor SNR on each branch.

(NOTE: Gi is the αi & αi is

the ri in the previous figure)

11
 Selection Combining (SC) (2)
Received signal in branch i: Note: If α is Rayleigh distributed
 α2 is exponentially distributed
 γi is exponentially distributed with
pdf
Instantaneous SNR on each branch i:
Eb/σ2 is the SNR without fading (σ2 = N0)

Average SNR on each branch:

The probability that a single branch has SNR less than some threshold γT (Rayleigh fading):

Probability that all M independent diversity branches receive signals are less than γT (i.e.,
outage):

(Assuming equal
average SNR for all
branches)
If we put γ instead of γT, we will get the CDF of
12
the output SNR γSEL of the selection combiner.
 Selection Combining (SC) (3)
Probability that SNR > γT for one or more branches:

Average SNR of the diversity combiner output signal:

There are alternate ways to

get this expression. Look
into the book of A.
Goldsmith.

13
 Selection Combining (SC) (4)
Thus the average SNR improvement offered by selection diversity as

 Thus, the average SNR gain increases with M, but not linearly. As the
number of branches increases, diversity gain suffers from diminishing returns.
 Selection diversity offers an average improvement in the link margin
without requiring additional transmitter power or sophisticated receiver
circuitry.
 It is easy to implement, however, it required the signals to be monitored
at a faster rate than the fading process.
 It is not an optimal diversity technique, because it does not use all of the M
possible branches simultaneously.
14
Selection Combining (SC) (5)
The biggest gain is obtained by going from no
diversity to two-branch diversity. Increasing the
number of diversity branches from two to three will give
much less gain than going from one to two, and in
general increasing M yields diminishing returns in terms
of the SNR gain.

15
 Selection Combining (SC) (5)
Example:
Find the outage probability of BPSK modulation at Pb = 10−3 for a Rayleigh
fading channel with SC diversity for M = 1 (no diversity), M = 2 and M = 3.
Assume equal average branch SNRs 15 dB.

Solution:
A BPSK modulated signal with γb = 7 dB has Pb = 10−3. Thus, we have, γT = 7 dB.

Substituting γT = 100.7 and ᴦ = 101.5 into outage equation, we get –

Pout = 0.1466 for M = 1
Pout = 0.0215 for M = 2,
Pout = 0.0031 for M = 3.

We see that each additional branch reduces outage probability by almost an order of
magnitude.

16
 Maximal Ratio Combining (MRC)

Principle: The voltage signals from each of the M diversity branches

are co-phased to provide coherent voltage addition.
 The signal voltage on the ith branch becomes

17
 Maximal Ratio Combining (MRC)

Resulting signal
applied to the detector:

Assuming each branch has the same average noise power E[ni2] = σ2 =
N0), total noise power NT becomes -

18
 Maximal Ratio Combining (MRC)
The SNR at the detector, γM:

γM becomes maximum when Gi = αi (for i = 1, 2, …, M),

which is MRC

 SNR out of the MRC = Sum of the SNRs in each branch

19
Maximal Ratio Combining (MRC)

Pout =

20
 Maximal Ratio Combining (MRC)

 The average received SNR in MRC is simply the sum of the individual
average received SNR from each branch

21
Equal Gain Combining (EGC)

22
 Equal Gain Combining (EGC)
 MRC requires knowledge of the time-varying SNR
on each branch to provide for the variable weighting
capability, which can be very difficult to measure.
 In such cases, the branch weights are all set to
unity, but the signals from each branch are co-phased
to provide equal gain combining diversity.
 This allows the receiver to exploit signals that are
simultaneously received on each branch.

23
 Equal Gain Combining (EGC)
SNR out of the equal gain combiner:

Average received SNR:

As αi is Rayleigh distributed,

24
 Equal Gain Combining (EGC)
The performance of EGC is superior to selection diversity, however,
marginally inferior to MRC.

25
 Performance Criteria (1)
1. Signaling rate/spectral efficiency: bps/Hz or bits per transmission.

2. Coding gain: It is the gain from code design that increases effective SNR.

3. Diversity gain: It is the increase in SNR due to some diversity scheme, or how much
the transmission power can be reduced when a diversity scheme is introduced, without a
performance loss. Diversity gain is usually expressed in decibels, and sometimes as a
power ratio. It is the slope of the error vs SNR curve.

 c is a scaling constant specific to

the modulation employed and the
nature of the channel
 γc (γc ≥ 1) denotes coding gain
 M is the diversity order of the
system
26
 Performance Criteria (2)
4. Array gain: Array gain is defined as the increase in SNR obtained by combining
multiple antenna elements compared to a single element. It is a result of coherent
combining of multiple signals, either from multiple transmitters or from multiple
receivers. Multiple antennas at the transmitter and receiver can give array gain by
precoding and Rx antenna combining.

The difference between array gain and diversity gain is that array gain gives an
increase in the average SNR, while diversity gain makes the probability density
function of the instantaneous SNR more concentrated around its average value. In
some cases, array and diversity gain can be achieved simultaneously, while in other
cases only one of the gains is achieved. For example, precoding can provide diversity
gain in addition to array gain. This requires more detailed channel knowledge, and
the precoding weights need to be updated more frequently to follow the fast fading.27
 Time Diversity
Ref. Book:
Fundamentals of Wireless Communications - David Tse

28
 Time Diversity: Core Concept
 Diversity over time can be obtained via coding and interleaving:
information is coded and the coded symbols are dispersed over time in
different coherence periods so that different parts of the codewords
experience independent fades.

 The simplest form of time diversity scheme is repetition coding: the

same information symbol is transmitted over several signal paths.

 While repetition coding achieves the maximum diversity gain, it is

usually quite wasteful of the degrees of freedom of the channel.

 More sophisticated schemes can increase the data rate and achieve a
coding gain along with the diversity gain.

29
 Time Diversity: Core Concept
Consider coherent scenario: the receiver has perfect knowledge of the channel
gains and can coherently combine the received signals in the diversity paths.
Time diversity is achieved by averaging the fading of the channel over time.
Typically, the channel coherence time is of the order of tens to hundreds of symbols,
and therefore the channel is highly correlated across consecutive symbols.
To ensure that the coded symbols are transmitted through independent or nearly
independent fading gains, interleaving of codewords is required.
We transmit a codeword x = [x1, x2,…, xL] of length L symbols and the received signal
is given by

Assuming ideal interleaving so that consecutive symbols xl are transmitted sufficiently

far apart in time, we can assume that the hl are independent.
The parameter L is commonly called the number of diversity branches.
The additive noises w1,…, wL are i.i.d. CN(0, N0) random variables.
30
Time Diversity: Core Concept

Tc
 The codewords are
transmitted over
consecutive symbols (top)
and interleaved (bottom).

 A deep fade will wipe

out the entire codeword
Four Symbols = One codeword (L = 4) in the former case, but
only one coded symbol
from each codeword in
the latter.

 In the latter case, each

codeword can still be
recovered from the other
three unfaded symbols. 31
 Time Diversity: Repetition Coding
The simplest code is a repetition code, in which xl = x1 for = 1,…, L.

In vector form, the overall channel becomes

Consider now coherent detection of x1, i.e., the channel gains are known to the receiver.

This is the canonical vector Gaussian detection problem. The below scalar is a sufficient statistic.

Thus, it is an equivalent scalar detection problem with noise (h∗/||h||) w ∼ CN(0, N0).

The receiver structure is a matched filter and is also called a maximal ratio combiner:
it weighs the received signal in each branch in proportion to the signal strength and also
aligns the phases of the signals in the summation to maximize the output SNR. This
receiver structure is also called coherent combining. 32
 Repetition Coding
For BPSK modulation, with x1 = ± A, the error probability, conditional on h, can be given by

Where SNR = A2/N0 = Eb/N0 is the average received signal-to-noise ratio per (complex) symbol
time, and ||h||2SNR is the received SNR for a given channel vector h. We average over ||h||2 to
find the overall error probability.

Under Rayleigh fading with each gain hl i.i.d. CN(0, N0),

It is a sum of the squares of 2L independent real Gaussian random variables, each term |hl|2
being the sum of the squares of the real and imaginary parts of hl. Therefore, ||h||2 is Chi-
square distributed with 2L degrees of freedom, and the density is given by

33
 Repetition Coding
The average error probability can be explicitly computed to be

At high SNR:

 In particular, the error probability decreases as the Lth power of SNR, corresponding to a
slope of −L in the error probability curve (in dB/dB scale).
 Typically, L is called the diversity gain of the system. 34
Repetition Coding

35
 Time Diversity: Beyond Repetition Coding
The repetition code is the simplest possible code. Although it achieves a
diversity gain, it does not exploit the degrees of freedom available in the
channel effectively, because it simply repeats the same symbol over the L
symbol times.

By using more sophisticated codes, a coding gain can also be obtained

beyond the diversity gain. There are many possible codes that one can use. We
will focus on the example of a rotation code to explain some of the issues in
code design for fading channels.

 Consider the case L = 2: A repetition code which repeats a BPSK symbol u

= ±a twice, obtains a diversity gain of 2, but would only transmit one bit
of information over the two symbol times.

Transmitting two independent BPSK symbols u1, u2 over the two times would
use the available degrees of freedom more efficiently, but of course offers no
diversity gain: an error would be made whenever one of the two channel
gains h1, h2 is in deep fade. 36
 Beyond Repetition Coding: Rotation Code
To get both benefits, consider instead a scheme that transmits the vector x over the
two symbol times (L = 2)
where

Here R is a rotation matrix (for some θ ∈ (0, 2π)). This is a code with four codewords:

1 = cos − sin
2 = sin + cos Fig. Codewords of
rotation code. 37
 Rotation Code
The received signal is given by

It is difficult to obtain an explicit expression for the exact error probability. So, we
will proceed by looking at the union bound.

Due to the symmetry of the code, without loss of generality we can assume xA is
transmitted. The union bound says that

where P{xA → xB} is the pairwise error probability of confusing xA with xB when xA is
transmitted and when these are the only two hypotheses. Conditioned on the channel
gains h1 and h2, this is just the binary detection problem with

38
 Rotation Code

where

d is the normalized difference between the codewords, normalized such that the
transmit energy is 1 per symbol time.

Averaging with respect to h1 and h2 under the independent Rayleigh fading assumption, we get

39
 Rotation Code
We first observe that if d1 = 0 or d2 = 0, then the diversity gain of the code is only 1.

If they are both non-zero, then at high SNR the above bound on the pairwise error
probability becomes

Here δAB is the squared product distance between xA and xB, when the average energy
of the code is normalized to be 1 per symbol time.

This δAB determines the pairwise error probability between the two codewords.
It is important that both |d1|2 and |d2|2 are large to ensure diversity against fading in
both components.

Similarly, we can define δij to be the squared product distance between xi and xj , i, j
= A, B, C, D.
40
 Rotation Code
Then the bound on the overall error probability:

41
 Repetition Coding: 4-PAM
(a) Codewords of
rotation code.

(b) Codewords
of repetition code.

It is interesting to see how this code compares to the repetition scheme.

To keep the bit rate the same (2 bits over 2 real-valued symbols), the repetition
scheme would be using 4-PAM modulation {−3b, −b, b, 3b}.

The codewords of the repetition scheme are shown in above figure. The pairwise
error probability between two adjacent codewords (say, xA and xB) is

42
 Repetition Coding: 4-PAM
But now SNR = 5b2/N0 is the average SNR per symbol time* for the 4-PAM
constellation, and d1 = d2 = 2/√5 are the normalized (such that the transmit energy is 1 per
symbol time, i.e., b = 1/√5) component differences between the adjacent codewords.

The minimum squared product distance for the repetition code is therefore 16/25
and we can compare this to the minimum squared product distance of 16/5 for the
previous rotation code.

Since the error probability is proportional to SNR−2 in both cases, we conclude that the
rotation code has an improved coding gain over the repetition code in terms of a
saving in transmit power by a factor of √5 (3.5 dB) for the same product distance.

This improvement comes from increasing the overall product distance, and this is
in turn due to spreading the codewords in the two-dimensional space rather than
packing them on a single-dimensional line as in the repetition code.

This is the same reason that QPSK is more efficient than BPSK.

*(Average energy per two symbol time = 10b2, per symbol time = 5b2) 43
 Repetition Coding: 4-PAM
The rotation code discussed above is specifically designed to exploit time diversity
in fading channels.

In the AWGN channel, however, rotation of the constellation does not affect
performance since the i.i.d. Gaussian noise is invariant to rotations.

On the other hand, codes that are designed for the AWGN channel, such as linear block
codes or convolutional codes, can be used to extract time diversity in fading channels
when combined with interleaving.

Nevertheless, these code design criteria all depend on the specific channel statistics
assumed.

44
 Frequency
Diversity
Ref. Book:
Fundamentals of Wireless Communications - David Tse

45
 What is Frequency Diversity?
So far we have focused on narrowband flat fading channels. These channels are
modeled by a single-tap filter, as most of the multi-paths arrive during one symbol
time.

In wideband channels, however, the transmitted signal arrives over multiple symbol
times and the multi-paths can be resolved at the receiver. The frequency response is no
longer flat, i.e., the transmission (signal) bandwidth W is greater than the channel
coherence bandwidth Wc of the channel. This provides frequency diversity.

46
 What is Frequency Diversity?
We begin with the discrete-time baseband model of the wireless channel. The sampled
output at time m denoted by y[m] can be written as

Here, hl[m] denotes the lth channel

filter tap at time m.

To understand the concept of frequency diversity in the simplest setting, consider first the
oneshot communication situation when one symbol x[0] is sent at time 0, and no symbol is
transmitted after that. The receiver observes

47
 What is Frequency Diversity?
If we assume that the channel response has a finite number of taps L, then the
delayed replicas of the signal are providing L branches of diversity in detecting
x[0], since the tap gains hl are assumed to be independent. This diversity is achieved
by the ability of resolving the multi-paths at the receiver due to the wideband nature
of the channel, and is thus called frequency diversity.

A simple communication scheme can be built on the above idea by sending an

information symbol every L symbol times.

The maximal diversity gain of L can be achieved, but the problem with this scheme is
that it is very wasteful of degrees of freedom: only one symbol can be transmitted
every delay spread (every L symbol times). This scheme can actually be thought of as
analogous to the repetition codes used for both time and spatial diversity, where one
information symbol is repeated L times.

In this setting, once one tries to transmit symbols more frequently, inter-symbol
interference (ISI) occurs: the delayed replicas of previous symbols interfere with the
current symbol.
48
 More on ISI

Suppose two sequences h[n] and X[n] of length L and P, respectively.

Then, Y[n] is a sequence of length (L + P − 1).

49
 What is Frequency Diversity?
The problem is then how to deal with the ISI while at the same time exploiting the
inherent frequency diversity in the channel.

Broadly speaking, there are three common approaches:

 Single-carrier systems with equalization
 Direct-sequence spread-spectrum (DSSS)
 Multi-carrier systems: OFDM

An important conceptual point is that, while frequency diversity is something

intrinsic in a wideband channel, the presence of ISI is not, as it depends on the
modulation technique used.

For example, under OFDM, there is no ISI, but sub-carriers that are separated by
more than the coherence bandwidth fade more or less independently and hence
frequency diversity is still present.

We assume that the receiver has a perfect estimate of the channel (CSI known).
50
 OFDM (Orthogonal frequency division multiplexing)
Both the single-carrier system with ISI equalization and the DS spread spectrum system
with Rake reception are based on a time-domain view of the channel.

But we know that if the channel is linear time-invariant (LTI), sinusoids are
eigenfunctions and they get transformed in a particularly simple way. ISI occurs in a
single-carrier system because the transmitted signals are not sinusoids. This suggests
that if the channel is underspread (i.e., the coherence time is much larger than the delay
spread) and is therefore approximately time-invariant for a sufficiently long time-scale,
then transformation into the frequency domain can be a fruitful approach for
communication over frequency-selective channels. This is the basic idea behind
OFDM.
Applications of OFDM/OFDMA:
LTE
WiMAX
DSL
ADSL
PLC
Digital TV
WLAN 51
 OFDM (Orthogonal frequency division multiplexing)
Basic idea of OFDM:

52
 OFDM
We begin with the discrete-time baseband model

For simplicity, we first assume that for each l, the lth tap is not changing with m and
hence the channel is linear time-invariant. Again assuming a finite number of non-
zero taps L = TdW (Td: delay spread), we can rewrite the channel model

Sinusoids are eigenfunctions of LTI systems, but they are of infinite duration. If we
transmit over only a finite duration, say Nc symbols, then the sinusoids are no longer
eigenfunctions. One way to restore the eigenfunction property is by adding a cyclic
prefix to the symbols. For every block of symbols of length Nc, denoted by

we create an Nc +L−1 input block as

53
 OFDM
That means, we add a prefix of length L−1 consisting of data symbols rotated
cyclically.

Time domain
data symbols
N = Nc
Time domain
OFDM symbol
Frequency domain
data symbols

Time domain
54
 OFDM

(L−1)
symbols Time domain

With this input to the channel, consider the output

The ISI extends over the first (L−1) symbols and the receiver ignores it by
considering only the output over the time interval m ∈ [L, Nc + L−1]. 55
 OFDM
Due to the additional cyclic prefix, the output over this time interval (of length Nc) is

m ∈ [L, Nc + L−1]

rotation

N = Nc

56
 OFDM
Denoting the output of length Nc by (time-domain)

Time domain

and the channel by a vector of length Nc

Time domain

Cyclic Prefix converts the linear convolution into a circular convolution

Thus, output y of length Nc can be written as (A)

Here, is a vector of i.i.d. CN(0, N0) random

variables.

The notation ⊗ denotes the cyclic / circular convolution.

Now, using the following identity,

57
 OFDM
Taking the discrete Fourier transform (DFT) of both sides of (A) , we can write in
frequency domain,

Also, DFT of the symbols,

58
 OFDM: Matrix Notation

Here, C is a circulant matrix, i.e., the rows are cyclic shifts of each other.

U is the unitary matrix (DFT matrix) with its (k, n)th entry equal to
U-1: Nc × Nc
IDFT matrix

Λ (Nc × Nc) is the diagonal matrix with diagonal entries √Nc times the DFT of h, i.e.,

59
 OFDM: Matrix Notation
Output y in equation (A) can thus be written as

Time domain

Time domain
Frequency Frequency
domain Domain
N = Nc

d y
h
U-1 U
60
 OFDM: Operation Summary
 The data symbols modulate Nc tones or sub-carriers, which occupy the bandwidth W and are
uniformly separated by W/Nc Hz.
 The data symbols on the sub-carriers are then converted (through the IDFT) to time domain.
 The procedure of introducing the cyclic prefix (CP) before transmission allows for the removal
of ISI. The receiver ignores the part of the output signal containing the CP (along with the ISI
terms) and converts the length Nc symbols back to the frequency domain through a DFT.
 The data symbols on the sub-carriers are maintained to be orthogonal as they propagate through
the channel and hence go through narrowband parallel sub-channels.
 DFT and IDFT can be very efficiently implemented (using FFT) whenever Nc is a power of 2.
Time domain
Frequency Frequency
domain Domain

d y
h
U-1 U
61
OFDM: Transmitter (including passband moduation)

Frequency Time
domain domain

62
OFDM: Receiver (including passband moduation)
Time Frequency
domain domain

63
 OFDM Block Length Nc (1)
The OFDM scheme converts communication over a multipath channel into
communication over simpler parallel narrowband sub-channels.
However, this simplicity is achieved at a cost of underutilizing two resources, resulting
in a loss of performance.
 First, the CP occupies an amount of time which cannot be used
to communicate data. This loss amounts to a fraction L/(Nc + L) of
the total time.
 The second loss is in the power transmitted. A fraction L/(Nc + L)
of the average power is allocated to the CP and cannot be used
towards communicating data.

(1) Thus, to minimize the overhead (in both time and power) due to the cyclic prefix,
we prefer to have Nc as large as possible. The time-varying nature of the wireless
channel, however, constrains the largest value Nc can reasonably take.
64
 OFDM Block Length Nc (2)
Consider a simple channel model that does not vary with time. If the channel is
slowly time varying, then the coherence time Tc is much larger than the delay spread Td
(the underspread scenario). For underspread channels, the block length of the OFDM
communication scheme Nc can be chosen significantly larger than the multipath length
L = TdW = Td/Ts (Ts = 1/W = sampling interval = symbol duration), but still much
smaller than the coherence block length TcW = Tc/Ts. That is, L << Nc < TcW. Under
these conditions, the channel model of linear time invariance approximates a slowly
time-varying channel over the block length Nc, while keeping the overhead small.

The constraint on the OFDM block length can also be understood in the frequency
domain. A block length of Nc corresponds to an inter-sub-carrier spacing equal to W/Nc.
In a wireless channel, the Doppler spread introduces uncertainty in the frequency of the
received signal; the Doppler spread is inversely proportional to the coherence time of
the channel: Ds = 1/4Tc. For the inter-sub-carrier spacing to be much larger than the
Doppler spread (i.e., W/Nc >> 1/4Tc), the OFDM block length Nc should be constrained
to be much smaller than TcW. This is the same constraint as above.

65
 OFDM Block Length Nc (3)
(2) Apart from an underutilization of time due to the presence of the CP,
as mentioned earlier that additional power is required due to the CP.

OFDM schemes that put a zero signal instead of the CP have been
proposed to reduce this loss. However, due to the abrupt transition in
the signal, such schemes introduce harmonics that are difficult to filter in
the overall signal.

Further, the CP can be used for timing and frequency acquisition in

wireless applications, and this capability would be lost if a zero signal
replaced the CP.

66
 OFDM: IEEE 802.11 (Wi-Fi) Case Study
The IEEE 802.11 Wireless LAN standard family (popularly known as Wi-Fi) consists
of a series of half-duplex over-the-air modulation techniques that use the same basic
protocol.

The 802.11 protocol family employs carrier-sense multiple access with collision
avoidance whereby equipment listens to a channel for other users (including non 802.11
users) before transmitting each packet.

The base version of the standard was released in 1997.

802.11-1997 was the first wireless networking standard in the family, but 802.11b was
the first widely accepted one, followed by 802.11a, 802.11g, 802.11n, and 802.11ac.

Other to a previous specification. standards in the family (c–f, h, j) are service

amendments that are used to extend the current scope of the existing standard, which
may also include corrections

Wi-Fi generations 1- 6 refer to the 802.11b, 802.11a, 802.11g, 802.11n, 802.11ac, and
67
802.11ax protocols, in that order.
 OFDM: 802.11 (Wi-Fi) Case Study
For example, the IEEE 802.11a Wireless LAN standards, which occupies 20 MHz of
bandwidth in the 5 GHz unlicensed band, is based on OFDM. The IEEE 802.11g
standard is virtually identical, but operates in the smaller and more crowded 2.4 GHz
unlicensed ISM band.

In 802.11a, N = 64 subcarriers are generated, although only 48 are actually used

for data transmission, with the outer 12 zeroed in order to reduce adjacent channel
interference, and 4 used as pilot symbols for channel estimation.

The CP consists of L-1 = 16 samples, so the total number of samples associated with
each OFDM symbol, including both data samples and the cyclic prefix, is 80. The
transmitter gets periodic feedback from the receiver about the packet error rate, which it
uses to pick an appropriate error correction code and modulation technique.

The same code and modulation must be used for all the subcarriers at any given time.
The error correction code is a convolutional code with one of three possible coding
rates: r = 1/2 , 2/3, or 3/4. The modulation types that can be used on the sub-channels
are BPSK, QPSK, 16QAM, or 64QAM.
68
 OFDM: Frequency Diversity
Let us revert to the non-overlapping narrowband channel representation of the ISI
channel as below:

The correlation between the channel frequency coefficients depends

on the coherence bandwidth of the channel. Now, coherence bandwidth is inversely
proportional to the multipath spread as shown below.

where we use our notation for L as denoting the length of the ISI. Since each sub-carrier
is W/Nc wide, we expect approximately

as the number of neighboring sub-carriers whose channel coefficients are heavily

correlated.
69
 OFDM: Frequency Diversity
One way to exploit the frequency diversity is to consider ideal interleaving across the
sub-carriers (analogous to the time-interleaving) and consider the model as before.

The difference is that l now represents the sub-carriers, while it is used to denote time in
time diversity case.

However, with the ideal frequency interleaving assumption, we retain the same
independent assumption on the channel coefficients. Thus, the discussion of time
diversity schemes is directly applicable here.

In particular, an L-fold diversity gain (proportional to the number of ISI symbols

L) can be obtained.

Since the communication scheme is over sub-carriers, the form of diversity is due
to the frequency-selective channel and is termed frequency diversity (as compared to
the time diversity which arises due to the time variations of the channel).
70
 OFDM: Frequency Diversity

Don’t experience any substantial

frequency diversity

Frequency diversity is achieved through

frequency interleaving of the coded bits
over the overall transmission bandwidth
of the OFDM signal.
71
The END

72
 OFDM

Frequency Time
domain Domain

OFDM transmission is block based,

implying that, during each OFDM
symbol interval, Nc modulation symbols
are transmitted in parallel.
73
 Additional Information (A. Goldsmith)
We make the standard assumption that w(t) is zero-mean additive white Gaussian noise (AWGN) with
power spectral density N0/2, i.e., E[w(0)w(t)] = N0/2 δ(t).

Hence the discrete-time noise process w(m) is white, i.e., independent over time; moreover, the real and
imaginary components are i.i.d. Gaussians with variances N0/2.

Complex noise, w ∼ CN (0, N0)

The noise energy per complex symbol time is N0

The assumption of AWGN essentially means that we are assuming that the primary source of the noise is at
the receiver or is radiation impinging on the receiver that is independent of the paths over which the signal is
being received. This is normally a very good assumption for most communication situations.
The orthogonal modulation scheme considered here uses only real symbols and hence transmits only on the
I channel. Hence it may seem more natural to define the SNR in terms of noise energy per real symbol, i.e.,
N0/2. However, later we will consider modulation schemes that use complex symbols and hence transmit on
both the I and Q channels. In order to be consistent throughout, we choose to define SNR this way.
Channel tap gains: For Rayleigh fading, all the taps are modeled as: h ∼ CN (0, σ2). Then the magnitude
|h| of each tap is Rayleigh distributed and |h|2 is exponentially distributed. For flat fading, only one tap in the
channel model.
Usually, for Rayleigh fading, h ∼ CN (0, 1), where we normalize the variance to be 1.
Recall that we normalized the channel gain such that E[h2] = 1. 74