Wireless Communication Notes PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 296

LECTURE NOTES

ON

WIRELESS COMMUNICATION

ld
AND
NETWORKS

or
W
TU
JN
ll
A
SYLLABUS

UNIT-I
The Cellular Concept-System Design Fundamentals: Introduction, Frequency Reuse, Channel
Assignment Strategies, Handoff Strategies- Prioritizing Handoffs, Practical Handoff Considerations,
Interference and system capacity — Co channel Interference and system capacity, Channel planning for
Wireless Systems, Adjacent Channel interference , Power Control for Reducing interference, Trunking
and Grade of Service, Improving Coverage & Capacity in Cellular Systems- Cell Splitting, Sectoring.

ld
UNIT—II
Mobile Radio Propagation: Large-Scale Path Loss: Introduction to Radio Wave Propagation, Free Space
Propagation Model, Relating Power to Electric Field, The Three Basic Propagation Mechanisms,
Reflection- Reflection from Dielectrics, Brewster Angle, Reflection from prefect conductors, Ground
Reflection (Two-Ray) Model, Diffraction-Fresnel Zone Geometry, Knife-edge Diffraction Model,

or
Multiple knife-edge Diffraction, Scattering, Outdoor Propagation Models- Longley-Ryce Model,
Okumura Model, Hata Model, PCS Extension to Hata Model, Walfisch and Bertoni Model, Wideband
PCS Microcell Model, lndoor Propagation Models-Partition losses (Same Floor), Partition losses between
Floors, Log-distance path loss model, Ericsson Multiple Breakpoint Model, Attenuation Factor Model,
Signal penetration into buildings, Ray Tracing and Site Specific Modeling.
.
UNIT —III

W
Mobile Radio Propagation: Small —Scale Fading and Multipath: Small Scale Multipath propagation-
Factors influencing small scale fading, Doppler shift, Impulse Response Model of a multipath channel-
Relationship between Bandwidth and Received power, Small-Scale Multipath Measurements-Direct RF
Pulse System, Spread Spectrum Sliding Correlator Channel Sounding, Frequency Domain Channels
Sounding, Parameters of Mobile Multipath Channels-Time Dispersion Parameters, Coherence Bandwidth,
Doppler Spread and Coherence Time, Types of Small-Scale Fading-Fading effects Due to Multipath Time
TU
Delay Spread, Flat fading, Frequency selective fading, Fading effects Due to Doppler Spread-Fast fading,
slow fading, Statistical Models for multipath Fading Channels-Clarke’s model for flat fading, spectral
shape due to Doppler spread in Clarke’s model, Simulation of Clarke and Gans Fading Model, Level
crossing and fading statistics, Two-ray Rayleigh Fading Model.

UNIT -IV
Equalization and Diversity: Introduction, Fundamentals of Equalization, Training A Generic Adaptive
JN

Equalizer, Equalizers in a communication Receiver, Linear Equalizers, Non-linear Equalization-Decision


Feedback Equalization (DFE), Maximum Likelihood Sequence Estimation (MLSE) Equalizer, Algorithms
for adaptive equalization-Zero Forcing Algorithm, Least Mean Square Algorithm, Recursive least squares
algorithm. Diversity Techniques-Derivation of selection Diversity improvement, Derivation of Maximal
Ratio Combining improvement, Practical Space Diversity Consideration-Selection Diversity, Feedback or
Scanning Diversity, Maximal Ratio Combining, Equal Gain Combining, Polarization Diversity,
Frequency Diversity, Time Diversity, RAKE Receiver.
UNIT -V
ll

Wireless Networks: Introduction to wireless Networks, Advantages and disadvantages of Wireless Local
Area Networks, WLAN Topologies, WLAN Standard IEEE 802.11 ,IEEE 802.11 Medium Access
Control, Comparision
A

of IEEE 802.11 a,b,g and n standards, IEEE 802.16 and its enhancements, Wireless PANs, Hiper Lan,
WLL.
TEXT BOOKS
1. Wireless Communications, Principles, Practice — Theodore, S.Rappaport, 2nd Ed., 2002, PHI.
2. Wireless Communications-Andrea Goldsmith, 2005 Cambridge University Press.
3. Mobile Cellular Communication — Gottapu Sasibhushana Rao, Pearson Education, 2012.
REFERENCE BOOKS

ld
1. Principles of Wireless Networks — Kaveh Pah Laven and P. Krishna Murthy, 2002, PE
2. Wireless Digital Communications — Kamilo Feher, 1999, PHI.
3. Wireless Communication and Networking — William Stallings, 2003,PHI.
4. Wireless Communication — Upen Dalal, Oxford Univ. Press

or
5. Wireless Communications and Networking — Vijay K. Gary, Elsevier.

W
TU
JN
ll
A
The Cellular Engineering
Fundamentals

ld
Introduction

or
In Chapter 1, we have seen that the technique of substituting a single high power
transmitter by several low power transmitters to support many users is the backbone
of the cellular concept. In practice, the following four parameters are most important
while considering the cellular issues: system capacity, quality of service, spectrum

W
efficiency and power management. Starting from the basic notion of a cell, we would
deal with these parameters in the context of cellular engineering in this chapter.

What is a Cell?
TU

The power of the radio signals transmitted by the BS decay as the signals travel
away from it. A minimum amount of signal strength (let us say, x dB) is needed in
order to be detected by the MS or mobile sets which may the hand-held personal
units or those installed in the vehicles. The region over which the signal strength
JN

lies above this threshold value x dB is known as the coverage area of a BS and
it must be a circular region, considering the BS to be isotropic radiator. Such a
circle, which gives this actual radio coverage, is called the foot print of a cell (in
reality, it is amorphous). It might so happen that either there may be an overlap
between any two such side by side circles or there might be a gap between the
ll
A

4
ld
or
Figure 3.1: Footprint of cells showing the overlaps and gaps.

W
coverage areas of two adjacent circles. This is shown in Figure 3.1. Such a circular
geometry, therefore, cannot serve as a regular shape to describe cells. We need a
regular shape for cellular design over a territory which can be served by 3 regular
polygons, namely, equilateral triangle, square and regular hexagon, which can cover
TU
the entire area without any overlap and gaps. Along with its regularity, a cell must
be designed such that it is most reliable too, i.e., it supports even the weakest mobile
with occurs at the edges of the cell. For any distance between the center and the
farthest point in the cell from it, a regular hexagon covers the maximum area. Hence
regular hexagonal geometry is used as the cells in mobile communication.
JN

Frequency Reuse

Frequency reuse, or, frequency planning, is a technique of reusing frequencies and


channels within a communication system to improve capacity and spectral efficiency.
Frequency reuse is one of the fundamental concepts on which commercial wireless
ll

systems are based that involve the partitioning of an RF radiating area into cells.
The increased capacity in a commercial wireless network, compared with a network
A

with a single transmitter, comes from the fact that the same radio frequency can be
reused in a different area for a completely different transmission.
Frequency reuse in mobile cellular systems means that frequencies allocated to

5
ld
or
Figure 3.2: Frequency reuse technique of a cellular system.

the service are reused in a regular pattern of cells, each covered by one base station.
The repeating regular pattern of cells is called cluster. Since each cell is designed

W
to use radio frequencies only within its boundaries, the same frequencies can be
reused in other cells not far away without interference, in another cluster. Such cells
are called ‘co-channel’ cells. The reuse of frequencies enables a cellular system to
handle a huge number of calls with a limited number of channels. Figure 3.2 shows
a frequency planning with cluster size of 7, showing the co-channels cells in different
TU

clusters by the same letter. The closest distance between the co-channel cells (in
different clusters) is determined by the choice of the cluster size and the layout of
the cell cluster. Consider a cellular system with S duplex channels available for
use and let N be the number of cells in a cluster. If each cell is allotted K duplex
JN

channels with all being allotted unique and disjoint channel groups we have S = KN
under normal circumstances. Now, if the cluster are repeated M times within the
total area, the total number of duplex channels, or, the total number of users in the
system would be T = MS = KMN . Clearly, if K and N remain constant, then

T ∝M (3.1)
ll

and, if T and K remain constant, then


1
A

N∝ . (3.2)
M
Hence the capacity gain achieved is directly proportional to the number of times
a cluster is repeated, as shown in (3.1), as well as, for a fixed cell size, small N

6
decreases the size of the cluster with in turn results in the increase of the number
of clusters (3.2) and hence the capacity. However for small N, co-channel cells are
located much closer and hence more interference. The value of N is determined by
calculating the amount of interference that can be tolerated for a sufficient quality
communication. Hence the smallest N having interference below the tolerated limit
is used. However, the cluster size N cannot take on any value and is given only by

ld
the following equation

N = i2 + ij + j 2 , i ≥ 0, j ≥ 0, (3.3)

or
where i and j are integer numbers.

Ex. 1: Find the relationship between any two nearest co-channel cell distance D
and the cluster size N.
Solution: For hexagonal

W
cells, it can be shown that the distance between two adjacent
cell centers = 3R, where R is the radius of any cell. The normalized co-channel
cell distance Dn can be calculated by traveling ’i’ cells in one direction and then
traveling ’j’ cells in anticlockwise 120 o of the primary direction. Using law of vector
addition,
TU
D2 2 2 o o 2
(3.4)
n = j cos (30 ) + (i + j sin(30 ))

which turns out to be


. √
Dn = i2 + ij + j 2 = N. (3.5)

Multiplying the actual distance 3R between two adjacent cells with it, we get
JN

√ √
D = Dn 3R = 3NR. (3.6)

Ex. 2: Find out the surface area of a regular hexagon with radius R, the surface
area of a large hexagon with radius D, and hence compute the total number of cells
in this large hexagon.
ll

Hint: In general, this large hexagon with radius D encompasses the center cluster of
N cells and one-third of the cells associated with six other peripheral large hexagons.
Thus, the answer must be N + 6( 3N ) = 3N .
A

7
Channel Assignment Strategies

With the rapid increase in number of mobile users, the mobile service providers
had to follow strategies which ensure the effective utilization of the limited radio
spectrum. With increased capacity and low interference being the prime objectives,
a frequency reuse scheme was helpful in achieving this objectives. A variety of

ld
channel assignment strategies have been followed to aid these objectives. Channel
assignment strategies are classified into two types: fixed and dynamic, as discussed
below.

or
Fixed Channel Assignment (FCA)

In fixed channel assignment strategy each cell is allocated a fixed number of voice
channels. Any communication within the cell can only be made with the designated

W
unused channels of that particular cell. Suppose if all the channels are occupied,
then the call is blocked and subscriber has to wait. This is simplest of the channel
assignment strategies as it requires very simple circuitry but provides worst channel
utilization. Later there was another approach in which the channels were borrowed
TU
from adjacent cell if all of its own designated channels were occupied. This was
named as borrowing strategy. In such cases the MSC supervises the borrowing pro-
cess and ensures that none of the calls in progress are interrupted.

Dynamic Channel Assignment (DCA)


JN

In dynamic channel assignment strategy channels are temporarily assigned for use
in cells for the duration of the call. Each time a call attempt is made from a cell the
corresponding BS requests a channel from MSC. The MSC then allocates a channel
to the requesting the BS. After the call is over the channel is returned and kept in
a central pool. To avoid co-channel interference any channel that in use in one cell
ll

can only be reassigned simultaneously to another cell in the system if the distance
between the two cells is larger than minimum reuse distance. When compared to the
A

FCA, DCA has reduced the likelihood of blocking and even increased the trunking
capacity of the network as all of the channels are available to all cells, i.e., good
quality of service. But this type of assignment strategy results in heavy load on
switching center at heavy traffic condition.

8
Ex. 3: A total of 33 MHz bandwidth is allocated to a FDD cellular system with
two 25 KHz simplex channels to provide full duplex voice and control channels.
Compute the number of channels available per cell if the system uses (i) 4 cell, (ii)
7 cell, and (iii) 8 cell reuse technique. Assume 1 MHz of spectrum is allocated to
control channels. Give a distribution of voice and control channels.

ld
Solution: One duplex channel = 2 x 25 = 50 kHz of spectrum. Hence the total
available duplex channels are = 33 MHz / 50 kHz = 660 in number. Among these
channels, 1 MHz / 50 kHz = 20 channels are kept as control channels.

or
(a) For N = 4, total channels per cell = 660/4 = 165.
Among these, voice channels are 160 and control channels are 5 in number.
(b) For N = 7, total channels per cell are 660/7 ≈ 94. Therefore, we have to go for
a more exact solution. We know that for this system, a total of 20 control channels

W
and a total of 640 voice channels are kept. Here, 6 cells can use 3 control channels
and the rest two can use 2 control channels each. On the other hand, 5 cells can use
92 voice channels and the rest two can use 90 voice channels each. Thus the total
solution for this case is:
6 x 3 + 1 x 2 = 20 control channels, and,
TU

5 x 92 + 2 x 90 = 640 voice channels.


This is one solution, there might exist other solutions too.
(c) The option N = 8 is not a valid option since it cannot satisfy equation (3.3) by
two integers i and j.
JN

Handoff Process

When a user moves from one cell to the other, to keep the communication between
the user pair, the user channel has to be shifted from one BS to the other without
interrupting the call, i.e., when a MS moves into another cell, while the conversation
ll

is still in progress, the MSC automatically transfers the call to a new FDD channel
without disturbing the conversation. This process is called as handoff. A schematic
A

diagram of handoff is given in Figure 3.3.


Processing of handoff is an important task in any cellular system. Handoffs
must be performed successfully and be imperceptible to the users. Once a signal

9
ld
or
Figure 3.3: Handoff scenario at two adjacent cell boundary.

W
level is set as the minimum acceptable for good voice quality (Prmin ), then a slightly
stronger level is chosen as the threshold (PrH )at which handoff has to be made, as
shown in Figure 3.4. A parameter, called power margin, defined as

∆ = Pr H − Pr min (3.7)
TU

is quite an important parameter during the handoff process since this margin ∆ can
neither be too large nor too small. If ∆ is too small, then there may not be enough
time to complete the handoff and the call might be lost even if the user crosses the
cell boundary.
JN

If ∆ is too high o the other hand, then MSC has to be burdened with unnecessary
handoffs. This is because MS may not intend to enter the other cell. Therefore ∆
should be judiciously chosen to ensure imperceptible handoffs and to meet other
objectives.

Factors Influencing Handoffs


ll

The following factors influence the entire handoff process:


A

(a) Transmitted power: as we know that the transmission power is different for dif-
ferent cells, the handoff threshold or the power margin varies from cell to cell.
(b) Received power: the received power mostly depends on the Line of Sight (LoS)
path between the user and the BS. Especially when the user is on the boundary of

10
ld
or
from i-th cell to j-th cell.
W
Figure 3.4: Handoff process associated with power levels, when the user is going

the two cells, the LoS path plays a critical role in handoffs and therefore the power
margin ∆ depends on the minimum received power value from cell to cell.
TU

(c) Area and shape of the cell: Apart from the power levels, the cell structure also
a plays an important role in the handoff process.
(d) Mobility of users: The number of mobile users entering or going out of a partic-
ular cell, also fixes the handoff strategy of a cell.
JN

To illustrate the reasons (c) and (d), let us consider a rectangular cell with sides R1
and R2 inclined at an angle θ with horizon, as shown in the Figure 3.5. Assume N1
users are having handoff in horizontal direction and N2 in vertical direction per unit
length.

The number of crossings along R1 side is : (N1cosθ + N2sinθ)R1 and the number of
crossings along R2 side is : (N1sinθ + N2cosθ)R2.
ll

Then the handoff rate λH can be written as


A

λH = (N1cosθ + N2sinθ)R1 + (N1sinθ + N2 cosθ)R2. (3.8)

11
ld
or
Figure 3.5: Handoff process with a rectangular cell inclined at an angle θ.

R1 by A and equating
R2
dλ H
to zero, we get
dR 1 W
Now, given the fixed area A = R1R2, we need to find λmin
H for a given θ. Replacing

N1sinθ + N2cosθ
R21 = A( N cosθ + N sinθ ).
1 2
(3.9)
TU
Similarly, for R2, we get
N1cosθ + N2sinθ
R2 = A( ). (3.10)
2
N 1sinθ + N 2cosθ
.
From the above equations, we have λH = 2 A(N1 N2 + (N 2 + N 2)cosθsinθ) which
1 2

means it it minimized at θ = 0o. Hence λmin
H = 2 AN 1 N 2 . Putting the value of θ
JN

R1 N1
in (3.9) or (3.10), we have = . This has two implications: (i) that handoff is
R2 N2

minimized if rectangular cell is aligned with X-Y axis, i.e., θ = 0o, and, (ii) that the
number of users crossing the cell boundary is inversely proportional to the dimension
of the other side of the cell. The above analysis has been carried out for a simple
square cell and it changes in more complicated way when we consider a hexagonal
ll

cell.

Handoffs In Different Generations


A

In 1G analog cellular systems, the signal strength measurements were made by


the BS and in turn supervised by the MSC. The handoffs in this generation can
be termed as Network Controlled Hand-Off (NCHO). The BS monitors the signal

12
strengths of voice channels to determine the relative positions of the subscriber.
The special receivers located on the BS are controlled by the MSC to monitor the
signal strengths of the users in the neighboring cells which appear to be in need
of handoff. Based on the information received from the special receivers the MSC
decides whether a handoff is required or not. The approximate time needed to make
a handoff successful was about 5-10 s. This requires the value of ∆ to be in the

ld
order of 6dB to 12dB.
In the 2G systems, the MSC was relieved from the entire operation. In this
generation, which started using the digital technology, handoff decisions were mobile

or
assisted and therefore it is called Mobile Assisted Hand-Off (MAHO). In MAHO,
the mobile center measures the power changes received from nearby base stations
and notifies the two BS. Accordingly the two BS communicate and channel transfer
occurs. As compared to 1G, the circuit complexity was increased here whereas the

W
delay in handoff was reduced to 1-5 s. The value of ∆ was in the order of 0-5 dB.
However, even this amount of delay could create a communication pause.
In the current 3G systems, the MS measures the power from adjacent BS and
automatically upgrades the channels to its nearer BS. Hence this can be termed as
TU
Mobile Controlled Hand-Off (MCHO). When compared to the other generations,
delay during handoff is only 100 ms and the value of ∆ is around 20 dBm. The
Quality Of Service (QoS) has improved a lot although the complexity of the circuitry
has further increased which is inevitable.
All these types of handoffs are usually termed as hard handoff as there is a shift
JN

in the channels involved. There is also another kind of handoff, called soft handoff,
as discussed below.
Handoff in CDMA: In spread spectrum cellular systems, the mobiles share the same
channels in every cell. The MSC evaluates the signal strengths received from different
BS for a single user and then shifts the user from one BS to the other without actually
ll

changing the channel. These types of handoffs are called as soft handoff as there is
no change in the channel.
A

13
Handoff Priority

While assigning channels using either FCA or DCA strategy, a guard channel concept
must be followed to facilitate the handoffs. This means, a fraction of total available
channels must be kept for handoff requests. But this would reduce the carried
traffic and only fewer channels can be assigned for the residual users of a cell. A

ld
good solution to avoid such a dead-lock is to use DCA with handoff priority (demand
based allocation).

A Few Practical Problems in Handoff Scenario

or
(a) Different speed of mobile users: with the increase of mobile users in urban areas,
microcells are introduced in the cells to increase the capacity (this will be discussed
later in this chapter). The users with high speed frequently crossing the micro-cells

W
become burdened to MSC as it has to take care of handoffs. Several schemes thus
have been designed to handle the simultaneous traffic of high speed and low speed
users while minimizing the handoff intervention from the MSC, one of them being
the ‘Umbrella Cell’ approach. This technique provides large area coverage to high
speed users while providing small area coverage to users traveling at low speed. By
TU

using different antenna heights and different power levels, it is possible to provide
larger and smaller cells at a same location. As illustrated in the Figure 3.6, umbrella
cell is co-located with few other microcells. The BS can measure the speed of the
user by its short term average signal strength over the RVC and decides which cell
JN

to handle that call. If the speed is less, then the corresponding microcell handles
the call so that there is good corner coverage. This approach assures that handoffs
are minimized for high speed users and provides additional microcell channels for
pedestrian users.
(b) Cell dragging problem: this is another practical problem in the urban area with
additional microcells. For example, consider there is a LOS path between the MS
ll

and BS1 while the user is in the cell covered by BS2. Since there is a LOS with the
BS1, the signal strength received from BS1 would be greater than that received from
A

BS2. However, since the user is in cell covered by BS2, handoff cannot take place
and as a result, it experiences a lot of interferences. This problem can be solved by
judiciously choosing the handoff threshold along with adjusting the coverage area.

14
(c) Inter-system handoff: if one user is leaving the coverage area of one MSC and is
entering the area of another MSC, then the call might be lost if there is no handoff in
this case too. Such a handoff is called inter-system handoff and in order to facilitate
this, mobiles usually have roaming facility.

Interference & System Capacity

ld
Susceptibility and interference problems associated with mobile communications
equipment are because of the problem of time congestion within the electromag-

or
netic spectrum. It is the limiting factor in the performance of cellular systems. This
interference can occur from clash with another mobile in the same cell or because
of a call in the adjacent cell. There can be interference between the base stations
operating at same frequency band or any other non-cellular system’s energy leaking

W
inadvertently into the frequency band of the cellular system. If there is an interfer-
ence in the voice channels, cross talk is heard will appear as noise between the users.
The interference in the control channels leads to missed and error calls because of
digital signaling. Interference is more severe in urban areas because of the greater
TU
RF noise and greater density of mobiles and base stations. The interference can be
divided into 2 parts: co-channel interference and adjacent channel interference.

Co-channel interference (CCI)

For the efficient use of available spectrum, it is necessary to reuse frequency band-
JN

width over relatively small geographical areas. However, increasing frequency reuse
also increases interference, which decreases system capacity and service quality. The
cells where the same set of frequencies is used are call co-channel cells. Co-channel
interference is the cross talk between two different radio transmitters using the same
radio frequency as is the case with the co-channel cells. The reasons of CCI can be
ll

because of either adverse weather conditions or poor frequency planning or overly-


crowded radio spectrum.
A

If the cell size and the power transmitted at the base stations are same then CCI
will become independent of the transmitted power and will depend on radius of the
cell (R) and the distance between the interfering co-channel cells (D). If D/R ratio
is increased, then the effective distance between the co-channel cells will increase

15
and interference will decrease. The parameter Q is called the frequency reuse ratio
and is related to the cluster size. For hexagonal geometry

Q = D/R = 3N. (3.11)

From the above equation, small of ‘Q’ means small value of cluster size ‘N’ and
increase in cellular capacity. But large ‘Q’ leads to decrease in system capacity

ld
but increase in transmission quality. Choosing the options is very careful for the
selection of ‘N’, the proof of which is given in the first section.
The Signal to Interference Ratio (SIR) for a mobile receiver which monitors the

or
forward channel can be calculated as

S S (3.12)
= . i0
I i=1 Ii

where i0 is the number of co-channel interfering cells, S is the desired signal power

W
from the baseband station and Ii is the interference power caused by the i-th interfer-
ing co-channel base station. In order to solve this equation from power calculations,
we need to look into the signal power characteristics. The average power in the
mobile radio channel decays as a power law of the distance of separation between
transmitter and receiver. The expression for the received power Pr at a distance d
TU

can be approximately calculated as


d
Pr = P0( )−n (3.13)
d0
and in the dB expression as
d
JN

Pr(dB) = P0(dB) − 10n log( ) (3.14)


d0

where P0 is the power received at a close-in reference point in the far field region at
a small distance do from the transmitting antenna, and ‘n’ is the path loss exponent.
Let us calculate the SIR for this system. If Di is the distance of the i-th interferer
from the mobile, the received power at a given mobile due to i-th interfering cell
ll

is proportional to (Di)−n (the value of ’n’ varies between 2 and 4 in urban cellular
systems).
A

Let us take that the path loss exponent is same throughout the coverage area
and the transmitted power be same, then SIR can be approximated as
. i0 n
IS R− i (3.15)
= −n
i=1 D

16
where the mobile is assumed to be located at R distance from the cell center. If
we consider only the first layer of interfering cells and we assume that the interfer-
ing base stations are equidistant from the reference base station and the distance
between the cell centers is ’D’ then the above equation can be converted as
n √ n
S (D/R) ( 3N ) (3.16)
= =
I i0 i0

ld
which is an approximate measure of the SIR. Subjective tests performed on AMPS
cellular system which uses FM and 30 kHz channels show that sufficient voice quality
can be obtained by SIR being greater than or equal to 18 dB. If we take n=4

or
, the value of ’N’ can be calculated as 6.49. Therefore minimum N is 7. The
above equations are based on hexagonal geometry and the distances from the closest
interfering cells can vary if different frequency reuse plans are used.
We can go for a more approximate calculation for co-channel SIR. This is the

W
example of a 7 cell reuse case. The mobile is at a distance of D-R from 2 closest
interfering cells and approximately D+R/2, D, D-R/2 and D+R distance from other
interfering cells in the first tier. Taking n = 4 in the above equation, SIR can be
approximately calculated as
TU

S R−4 (3.17)
=
I 2(D − R)−4 + (D + R)−4 + (D)−4 + (D + R/2)−4 + (D − R/2)−4
which can be rewritten in terms frequency reuse ratio Q as

S 1 . (3.18)
=
I 2(Q − 1) + (Q + 1) + (Q) + (Q + 1/2)−4 + (Q − 1/2)−4
−4 −4 −4
JN

Using the value of N equal to 7 (this means Q = 4.6), the above expression yields
that worst case SIR is 53.70 (17.3 dB). This shows that for a 7 cell reuse case the
worst case SIR is slightly less than 18 dB. The worst case is when the mobile is at
the corner of the cell i.e., on a vertex as shown in the Figure 3.6. Therefore N = 12
cluster size should be used. But this reduces the capacity by 7/12 times. Therefore,
ll

co-channel interference controls link performance, which in a way controls frequency


reuse plan and the overall capacity of the cellular system. The effect of co-channel
A

interference can be minimized by optimizing the frequency assignments of the base


stations and their transmit powers. Tilting the base-station antenna to limit the
spread of the signals in the system can also be done.

17
ld
or
W
Figure 3.6: First tier of co-channel interfering cells

Adjacent Channel Interference (ACI)

This is a different type of interference which is caused by adjacent channels i.e.


TU
channels in adjacent cells. It is the signal impairment which occurs to one frequency
due to presence of another signal on a nearby frequency. This occurs when imperfect
receiver filters allow nearby frequencies to leak into the passband. This problem is
enhanced if the adjacent channel user is transmitting in a close range compared to
the subscriber’s receiver while the receiver attempts to receive a base station on the
JN

channel. This is called near-far effect. The more adjacent channels are packed into
the channel block, the higher the spectral efficiency, provided that the performance
degradation can be tolerated in the system link budget. This effect can also occur
if a mobile close to a base station transmits on a channel close to one being used
by a weak mobile. This problem might occur if the base station has problem in
ll

discriminating the mobile user from the ”bleed over” caused by the close adjacent
channel mobile.
A

Adjacent channel interference occurs more frequently in small cell clusters and heav-
ily used cells. If the frequency separation between the channels is kept large this
interference can be reduced to some extent. Thus assignment of channels is given

18
such that they do not form a contiguous band of frequencies within a particular
cell and frequency separation is maximized. Efficient assignment strategies are very
much important in making the interference as less as possible. If the frequency fac-
tor is small then distance between the adjacent channels cannot put the interference
level within tolerance limits. If a mobile is 10 times close to the base station than
other mobile and has energy spill out of its passband, then SIR for weak mobile is

ld
approximately
S
= 10−n (3.19)
I
which can be easily found from the earlier SIR expressions. If n = 4, then SIR is

or
−52 dB. Perfect base station filters are needed when close-in and distant users share
the same cell. Practically, each base station receiver is preceded by a high Q cavity
filter in order to remove adjacent channel interference. Power control is also very

W
much important for the prolonging of the battery life for the subscriber unit but also
reduces reverse channel SIR in the system. Power control is done such that each
mobile transmits the lowest power required to maintain a good quality link on the
reverse channel.
TU

Enhancing Capacity And Cell Coverage

The Key Trade-off

Previously, we have seen that the frequency reuse technique in cellular systems
JN

allows for almost boundless expansion of geographical area and the number of mobile
system users who could be accommodated. In designing a cellular layout, the two
parameters which are of great significance are the cell radius R and the cluster size

N, and we have also seen that co-channel cell distance D = 3NR. In the following,
a brief description of the design trade-off is given, in which the above two parameters
play a crucial role.
ll

The cell radius governs both the geographical area covered by a cell and also
the number of subscribers who can be serviced, given the subscriber density. It is
A

easy to see that the cell radius must be as large as possible. This is because, every
cell requires an investment in a tower, land on which the tower is placed, and radio
transmission equipment and so a large cell size minimizes the cost per subscriber.

19
Eventually, the cell radius is determined by the requirement that adequate signal
to noise ratio be maintained over the coverage area. The SNR is determined by
several factors such as the antenna height, transmitter power, receiver noise figure
etc. Given a cell radius R and a cluster size N , the geographic area covered by a
cluster is

Acluster = N Acell = N 3 3R2 /2. (3.20)

ld
If the total serviced area is Atotal, then the number of clusters M that could be
accommodated is given by

or
M = Atotal /Acluster = Atotal /(N 3 3R2 /2). (3.21)

Note that all of the available channels N, are reused in every cluster. Hence, to make
the maximum number of channels available to subscribers, the number of clusters

W
M should be large, which, by Equation (3.21), shows that the cell radius should
be small. However, cell radius is determined by a trade-off: R should be as large
as possible to minimize the cost of the installation per subscriber, but R should
be as small as possible to maximize the number of customers that the system can
accommodate. Now, if the cell radius R is fixed, then the number of clusters could be
TU

maximized by minimizing the size of a cluster N . We have seen earlier that the size
of a cluster depends on the frequency reuse ratio Q. Hence, in determining the value
of N , another trade-off is encountered in that N must be small to accommodate
large number of subscribers, but should be sufficiently large so as to minimize the
JN

interference effects.
Now, we focus on the issues regarding system expansion. The history of cellular
phones has been characterized by a rapid growth and expansion in cell subscribers.
Though a cellular system can be expanded by simply adding cells to the geographical
area, the way in which user density can be increased is also important to look at.
This is because it is not always possible to counter the increasing demand for cellular
ll

systems just by increasing the geographical coverage area due to the limitations in
obtaining new land with suitable requirements. We discuss here two methods for
A

dealing with an increasing subscriber density: Cell Splitting and Sectoring. The
other method, microcell zone concept can treated as enhancing the QoS in a cellular
system.

20
The basic idea of adopting the cellular approach is to allow space for the growth
of mobile users. When a new system is deployed, the demand for it is fairly low and
users are assumed to be uniformly distributed over the service area. However, as new
users subscribe to the cellular service, the demand for channels may begin to exceed
the capacity of some base stations. As discussed previously,the number of channels
available to customers (equivalently, the channel density per square kilometer) could

ld
be increased by decreasing the cluster size. However, once a system has been initially
deployed, a system-wide reduction in cluster size may not be necessary since user
density does not grow uniformly in all parts of the geographical area. It might be

or
that an increase in channel density is required only in specific parts of the system
to support an increased demand in those areas. Cell-splitting is a technique which
has the capability to add new smaller cells in specific areas of the system.

Cell-Splitting

W
Cell Splitting is based on the cell radius reduction and minimizes the need to modify
the existing cell parameters. Cell splitting involves the process of sub-dividing a
congested cell into smaller cells, each with its own base station and a corresponding
TU
reduction in antenna size and transmitting power. This increases the capacity of
a cellular system since it increases the number of times that channels are reused.
Since the new cells have smaller radii than the existing cells, inserting these smaller
cells, known as microcells, between the already existing cells results in an increase
of capacity due to the additional number of channels per unit area. There are few
JN

challenges in increasing the capacity by reducing the cell radius. Clearly, if cells
are small, there would have to be more of them and so additional base stations
will be needed in the system. The challenge in this case is to introduce the new
base stations without the need to move the already existing base station towers.
The other challenge is to meet the generally increasing demand that may vary quite
ll

rapidly between geographical areas of the system. For instance, a city may have
highly populated areas and so the demand must be supported by cells with the
A

smallest radius. The radius of cells will generally increase as we move from urban to
sub urban areas, because the user density decreases on moving towards sub-urban
areas. The key factor is to add as minimum number of smaller cells as possible

21
ld
or
W
Figure 3.7: Splitting of congested seven-cell clusters.

wherever an increase in demand occurs. The gradual addition of the smaller cells
implies that, at least for a time, the cellular system operates with cells of more than
TU
one size.
Figure 3.7 shows a cellular layout with seven-cell clusters. Consider that the cells
in the center of the diagram are becoming congested, and cell A in the center has
reached its maximum capacity. Figure also shows how the smaller cells are being
superimposed on the original layout. The new smaller cells have half the cell radius
JN

of the original cells. At half the radius, the new cells will have one-fourth of the area
and will consequently need to support one-fourth the number of subscribers. Notice
that one of the new smaller cells lies in the center of each of the larger cells. If
we assume that base stations are located in the cell centers, this allows the original
base stations to be maintained even in the new system layout. However, new base
ll

stations will have to be added for new cells that do not lie in the center of the larger
cells. The organization of cells into clusters is independent of the cell radius, so that
A

the cluster size can be the same in the small-cell layout as it was in the large-cell
layout. Also the signal-to-interference ratio is determined by cluster size and not by
cell radius. Consequently, if the cluster size is maintained, the signal-to-interference
ratio will be the same after cell splitting as it was before. If the entire system is

22
replaced with new half-radius cells, and the cluster size is maintained, the number
of channels per cell will be exactly as it was before, and the number of subscribers
per cell will have been reduced.
When the cell radius is reduced by a factor, it is also desirable to reduce the
transmitted power. The transmit power of the new cells with radius half that of the
old cells can be found by examining the received power PR at the new and old cell

ld
boundaries and setting them equal. This is necessary to maintain the same frequency
re-use plan in the new cell layout as well. Assume that PT1 and PT2 are the transmit
powers of the larger and smaller base stations respectively. Then, assuming a path

or
loss index n=4, we have power received at old cell boundary = PT 1/R4 and the
power received at new cell boundary = PT 2 /(R/2)4 . On equating the two received
powers, we get PT 2 = PT 1 / 16. In other words, the transmit power must be reduced
by 12 dB in order to maintain the same S/I with the new system lay-out.

W
At the beginning of this channel splitting process, there would be fewer channels
in the smaller power groups. As the demand increases, more and more channels need
to be accommodated and hence the splitting process continues until all the larger
cells have been replaced by the smaller cells, at which point splitting is complete
TU
within the region and the entire system is rescaled to have a smaller radius per cell.
If a cellular layout is replaced entirety by a new layout with a smaller cell radius, the
signal-to-interference ratio will not change, provided the cluster size does not
change. Some special care must be taken, however, to avoid co-channel interference
when both large and small cell radii coexist. It turns out that the only way to
JN

avoid interference between the large-cell and small-cell systems is to assign entirely
different sets of channels to the two systems. So, when two sizes of cells co-exist in
a system, channels in the old cell must be broken down into two groups, one that
corresponds to larger cell reuse requirements and the other which corresponds to the
smaller cell reuse requirements. The larger cell is usually dedicated to high speed
ll

users as in the umbrella cell approach so as to minimize the number of hand-offs.

Ex. 4: When the AMPS cellular system was first deployed, the aim of the
A

system designers was to guarantee coverage. Initially the number of users was not
significant. Consequently cells were configured with an eight-mile radius, and a 12-
cell cluster size was chosen. The cell radius was chosen to guarantee a 17 dB

23
ld
Figure 3.8: A cell divided into three 120 o sectors.

or
signal-to-noise ratio over 90% of the coverage area. Although a 12-cell cluster size
provided more than adequate co-channel separation to meet a requirement for a
17 dB signal-to-interference ratio in an interference-limited environment, it did not

W
provide adequate frequency reuse to service an explosively growing customer base.
The system planners reasoned that a subsequent shift to a 7-cell cluster size would
provide an adequate number of channels. It was estimated that a 7-cell cluster
size should provide an adequate 18.7 dB signal-to-interference ratio. The margin,
TU
however, is slim, and the 17 dB signal-to-interference ratio requirement could not
be met over 90 % of the coverage area.

Sectoring

Sectoring is basically a technique which can increase the SIR without necessitating
JN

an increase in the cluster size. Till now, it has been assumed that the base station is
located in the center of a cell and radiates uniformly in all the directions behaving as
an omni-directional antenna. However it has been found that the co-channel inter-
ference in a cellular system may be decreased by replacing a single omni-directional
antenna at the base station by several directional antennas, each radiating within a
ll

specified sector. In the Figure 3.8, a cell is shown which has been split into three
120o sectors. The base station feeds three 120o directional antennas, each of which
A

radiates into one of the three sectors. The channel set serving this cell has also been
divided, so that each sector is assigned one-third of the available number cell of
channels. This technique for reducing co-channel interference wherein by using suit-

24
ld
Figure 3.9: A seven-cell cluster with 60o sectors.

or
able directional antennas, a given cell would receive interference and transmit with
a fraction of available co-channel cells is called ’sectoring’. In a seven-cell-cluster
layout with 120o sectored cells, it can be easily understood that the mobile units in

W
a particular sector of the center cell will receive co-channel interference from only
two of the first-tier co-channel base stations, rather than from all six. Likewise, the
base station in the center cell will receive co-channel interference from mobile units
in only two of the co-channel cells. Hence the signal to interference ratio is now
TU
modified to √ n
S ( 3N ) (3.22)
I = 2
where the denominator has been reduced from 6 to 2 to account for the reduced
number of interfering sources. Now, the signal to interference ratio for a seven-cell
cluster layout using 120o sectored antennas can be found from equation (3.24) to be
JN

23.4 dB which is a significant improvement over the Omni-directional case where the
worst-case S/I is found to be 17 dB (assuming a path-loss exponent, n=4). Some
cellular systems divide the cells into 60o sectors. Similar analysis can be performed
on them as well.
Ex. 5: A cellular system having a seven-cell cluster layout with omni-directional
ll

antennas has been performing satisfactorily for a required signal to interference ratio
of 15 dB. However due to the need for increasing the number of available channels, a
A

60o sectoring of the cells has been introduced. By what percentage can the number
of channels Ntotal be increased assuming a path-loss component n=4?
Solution: The seven-cell cluster layout with 60o sectoring is shown in the Figure 3.9.

25
It is easy to see that the shaded region in the center receives interference from just
one first-tier cell and hence the signal to interference ratio can be obtained suitably
as
√ n
, 4
S ( 3N ) ( (3)(7)) = 26.4dB. (3.23)
I = 1 = 1
Since the SIR exceeds 15 dB, one can try reducing the cluster size from seven to

ld
four. Now, the SIR for this reduced cluster size layout can be found to be
√ n
, 4
S ( 3N ) ( (3)(4)) = 21.6dB. (3.24)
I = 1 = 1

The S/I ratio is still above the requirement and so a further reduction in the cell

or
cluster size is possible. For a 3-cell cluster layout, there are two interfering sources
and hence the S/I ratio is found to be
√ √
3N )
n
( 33)4
S ( = = 16.07dB. (3.25)
I = 1

W 2

This is just above the adequate S/I ratio and further reduction in cluster size is
not possible. So, a 3-cluster cell layout could be used for meeting the growth re-
quirements. Thus, when the cluster size is reduced from 7 to 3, the total number of
TU

channels increased by a factor of 7/3.


The calculations in the above example are actually an idealization for several
reasons. Firstly, practical antennas have side lobes and cannot be used to focus a
transmitted beam into a perfect 120o sector or 60o sector. Due to this, additional
JN

interference will be introduced. Next, it is also a cause of concern that a given


number of channels are not able to support as many subscribers when the pool
of channels is divided into small groups. This is due to a reduction in Trunking
Efficiency, a term which will be explained later on. Because sectoring involves using
more than one antenna per base station, the available channels in the cell are divided
and dedicated to a specific antenna. This breaks the available set of channels into
ll

smaller sets, thus reducing the trunking efficiency. Moreover, dividing a cell into
sectors requires that a call in progress will have to be handed off (that is, assigned
A

a new channel) when a mobile unit travels into a new sector. This increases the
complexity of the system and also the load on the mobile switching center/base
station.

26
3.7.4 Microcell Zone Concept

The increased number of handoffs required when sectoring is employed results in an


increased load on the switching and control link elements of the mobile system. To
overcome this problem, a new microcell zone concept has been proposed. As shown
in Figure 3.10, this scheme has a cell divided into three microcell zones, with each

ld
of the three zone sites connected to the base station and sharing the same radio
equipment. It is necessary to note that all the microcell zones, within a cell, use the
same frequency used by that cell; that is no handovers occur between microcells.
Thus when a mobile user moves between two microcell zones of the cell, the BS

or
simply switches the channel to a different zone site and no physical re-allotment of
channel takes place.
Locating the mobile unit within the cell: An active mobile unit sends a signal to all

strongest signal. W
zone sites, which in turn send a signal to the BS. A zone selector at the BS uses that
signal to select a suitable zone to serve the mobile unit - choosing the zone with the

Base Station Signals: When a call is made to a cellular phone, the system already
knows the cell location of that phone. The base station of that cell knows in which
TU

zone, within that cell, the cellular phone is located. Therefore when it receives the
signal, the base station transmits it to the suitable zone site. The zone site receives
the cellular signal from the base station and transmits that signal to the mobile
phone after amplification. By confining the power transmitted to the mobile phone,
JN

co-channel interference is reduced between the zones and the capacity of system is
increased.
Benefits of the micro-cell zone concept: 1) Interference is reduced in this case as
compared to the scheme in which the cell size is reduced.
2) Handoffs are reduced (also compared to decreasing the cell size) since the micro-
cells within the cell operate at the same frequency; no handover occurs when the
ll

mobile unit moves between the microcells.


3) Size of the zone apparatus is small. The zone site equipment being small can be
A

mounted on the side of a building or on poles.


4) System capacity is increased. The new microcell knows where to locate the mo-
bile unit in a particular zone of the cell and deliver the power to that zone. Since

27
ld
or
W
Figure 3.10: The micro-cell zone concept.

the signal power is reduced, the microcells can be closer and result in an increased
TU

system capacity. However, in a microcellular system, the transmitted power to a


mobile phone within a microcell has to be precise; too much power results in inter-
ference between microcells, while with too little power the signal might not reach
the mobile phone.This is a drawback of microcellular systems, since a change in the
JN

surrounding (a new building, say, within a microcell) will require a change of the
transmission power.

Trunked Radio System

In the previous sections, we have discussed the frequency reuse plan, the design
ll

trade-offs and also explored certain capacity expansion techniques like cell-splitting
and sectoring. Now, we look at the relation between the number of radio channels
A

a cell contains and the number of users a cell can support. Cellular systems use
the concept of trunking to accommodate a large number of users in a limited radio
spectrum. It was found that a central office associated with say, 10,000 telephones

28
requires about 50 million connections to connect every possible pair of users. How-
ever, a worst case maximum of 5000 connections need to be made among these
telephones at any given instant of time, as against the possible 50 million connec-
tions. In fact, only a few hundreds of lines are needed owing to the relatively short
duration of a call. This indicates that the resources are shared so that the number of
lines is much smaller than the number of possible connections. A line that connects

ld
switching offices and that is shared among users on an as-needed basis is called a
trunk.
The fact that the number of trunks needed to make connections between offices

or
is much smaller than the maximum number that could be used suggests that at
times there might not be sufficient facilities to allow a call to be completed. A call
that cannot be completed owing to a lack of resources is said to be blocked. So one
important to be answered in mobile cellular systems is: How many channels per cell

a call will be blocked? W


are needed in a cellular telephone system to ensure a reasonably low probability that

In a trunked radio system, a channel is allotted on per call basis. The perfor-
mance of a radio system can be estimated in a way by looking at how efficiently the
TU
calls are getting connected and also how they are being maintained at handoffs.
Some of the important factors to take into consideration are (i) Arrival statistics,
(ii)Service statistics, (iii)Number of servers/channels.
Let us now consider the following assumptions for a bufferless system handling ’L’
users as shown in Figure 3.11:
JN

(i) The number of users L is large when compared to 1.


(ii) Arrival statistics is Poisson distributed with a mean parameter λ.

(iii) Duration of a call is exponentially distributed with a mean rate µ 1.


(iv) Residence time of each user is exponentially distributed with a rate parameter

µ 2.
ll

(v) The channel holding rate therefore is exponentially distributed with a parameter
µ = µ1 + µ2.
A

(vi) There is a total of ’J’ number of channels (J ≤ L).


To analyze such a system, let us recapitulate a queuing system in brief. Consider an
M/M/m/m system which is an m-server loss system. The name M/M/m/m reflects

29
ld
or
W
Figure 3.11: The bufferless J-channel trunked radio system.
TU
JN
ll

Figure 3.12: Discrete-time Markov chain for the M/M/J/J trunked radio system.
A

30
standard queuing theory nomenclature whereby:
(i) the first letter indicates the nature of arrival process(e.g. M stands for memory-
less which here means a Poisson process).
(ii) the second letter indicates the nature of probability distribution of service
times.(e.g M stands for exponential distribution). In all cases,successive inter ar-
rival times and service times are assumed to be statistically independent of each

ld
other.
(iii) the third letter indicates the number of servers.
(iv) the last letter indicates that if an arrival finds all ’m’ users to be busy, then it

or
will not enter the system and is lost.
In view of the above, the bufferless system as shown in Figure 3.11 can be modeled
as M/M/J/J system and the discrete-time Markov chain of this system is shown in
Figure 3.12.

W
Trunking mainly exploits the statistical behavior of users so that a fixed number
of channels can be used to accommodate a large, random user community. As the
number of telephone lines decrease, it becomes more likely that all channels are
busy for a particular user. As a result, the call gets rejected and in some systems,
TU
a queue may be used to hold the caller’s request until a channel becomes available.
In the telephone system context the term Grade of Service (GoS) is used to mean
the probability that a user’s request for service will be blocked because a required
facility, such as a trunk or a cellular channel, is not available. For example, a GoS of
2 % implies that on the average a user might not be successful in placing a call on 2
JN

out of every 100 attempts. In practice the blocking frequency varies with time. One
would expect far more call attempts during business hours than during the middle of
the night. Telephone operating companies maintain usage records and can identify a
”busy hour”, that is, the hour of the day during which there is the greatest demand
for service. Typically, telephone systems are engineered to provide a specified grade
ll

of service during a specified busy hour.


User calling can be modeled statistically by two parameters: the average number
A

of call requests per unit time λuser and the average holding time H. The parameter
λuser is also called the average arrival rate, referring to the rate at which calls from
a single user arrive. The average holding time is the average duration of a call. The

31
product:
Auser = λuser H (3.26)

that is, the product of the average arrival rate and the average holding time–is called
the offered traffic intensity or offered load. This quantity represents the average
traffic that a user provides to the system. Offered traffic intensity is a quantity that
is traditionally measured in Erlangs. One Erlang represents the amount of traffic

ld
intensity carried by a channel that is completely occupied. For example, a channel
that is occupied for thirty minutes during an hour carries 0.5 Erlang of traffic.
Call arrivals or requests for service are modeled as a Poisson random process. It

or
is based on the assumption that there is a large pool of users who do not cooperate
in deciding when to place calls. Holding times are very well predicted using an
exponential probability distribution. This implies that calls of long duration are

W
much less frequent than short calls. If the traffic intensity offered by a single user is
Auser , then the traffic intensity offered by N users is A = NAuser . The purpose of
the statistical model is to relate the offered traffic intensity A, the grade of service
Pb, and the number of channels or trunks C needed to maintain the desired grade
of service.
TU
Two models are widely used in traffic engineering to represent what happens
when a call is blocked. The blocked calls cleared model assumes that when a channel
or trunk is not available to service an arriving call, the call is cleared from the
system. The second model is known as blocked calls delayed. In this model a call
that cannot be serviced is placed on a queue and will be serviced when a channel or
JN

trunk becomes available.


Use of the blocked-calls-cleared statistical model leads to the Erlang B formula
that relates offered traffic intensity A, grade of service Pb, and number of channels
K. The Erlang B formula is:
AK /K!
Pb = . (3.27)
ll

K n /n!
n=0 A
When the blocked-calls-delayed model is used, the ”grade of service” refers to the
A

probability that a call will be delayed. In this case the statistical model leads to the
Erlang C formula,
AK /[(K A)(K 1)]!
P [delay] = − − . (3.28)
.
A /[(K − A)(K − 1)]! + K
K
n=0 A n /n!

32
Ex. 6: In a certain cellular system, an average subscriber places two calls per
hour during a busy hour and the average holding time is 3 min. Each cell has 100
channels. If the blocked calls are cleared, how many subscribers can be serviced by
each cell at 2 % GoS?
Solution: Using Erlang B table, it can be seen that for C = 100 and ttoS = Pb = 2%,

ld
the total offered load A=87.972 Erlangs. Since an individual subscriber offers a load
of Auser = (2 calls / 60 min)3 min = 0.1 Erlang, the maximum number of subscribers
served is

or
N = A/Auser = 87.972/0.1 ≈ 880. (3.29)

Ex. 4: In the previous example, suppose that the channels have been divided into
two groups of 50 channels each. Each subscriber is assigned to a group and can be
served only by that group. How many subscribers can be served by the two group
cell?

offered load per group is


W
Solution: Using the Erlang B table with C = 50 and ttOS = Pb = 2%, the total

A = 40.255Erlangs (3.30)
TU
Thus the maximum number of users per group is

Ngroup = A/Auser ≈ 403. (3.31)

Thus, counting both the groups, maximum number of users in the two group cell is
JN

806.
The above example indicates that the number of subscribers that can be sup-
ported by a given number of channels decreases as the pool of channels is sub-divided.
We can express this in terms of the trunking efficiency, defined as the carrier load
per channel, that is,

ξ = (1 − Pb )A/C. (3.32)
ll

This explains why the sectoring of a cell into either 120o or 60o sectors reduces
the trunking efficiency of the system. Thus the system growth due to sectoring is
A

impacted by trunking efficiency considerations.

33
References

1. T. S. Rappaport, Wireless Communications: Principles and Practice, 2nd ed.


Singapore: Pearson Education, Inc., 2002.

2. K. Feher, Wireless Digital Communications: Modulation and Spread Spectrum


Applications. Upper Saddle River, NJ: Prentice Hall, 1995.

ld
3. S. Haykin and M. Moher, Modern Wireless Communications. Singapore: Pear-
son Education, Inc., 2002.

or
4. J. W. Mark and W. Zhuang, Wireless Communications and Networking. New
Delhi: PHI, 2005.

W
TU
JN
ll
A

34
Chapter 4

ld
Free Space Radio Wave

or
Propagation

Introduction
W
There are two basic ways of transmitting an electro-magnetic (EM) signal, through a
guided medium or through an unguided medium. Guided mediums such as coaxial
cables and fiber optic cables, are far less hostile toward the information carrying
TU

EM signal than the wireless or the unguided medium. It presents challenges and
conditions which are unique for this kind of transmissions. A signal, as it travels
through the wireless channel, undergoes many kinds of propagation effects such as
reflection, diffraction and scattering, due to the presence of buildings, mountains and
JN

other such obstructions. Reflection occurs when the EM waves impinge on objects
which are much greater than the wavelength of the traveling wave. Diffraction
is a phenomena occurring when the wave interacts with a surface having sharp
irregularities. Scattering occurs when the medium through the wave is traveling
contains objects which are much smaller than the wavelength of the EM wave.
These varied phenomena’s lead to large scale and small scale propagation losses. Due
ll

to the inherent randomness associated with such channels they are best described
with the help of statistical models. Models which predict the mean signal strength
A

for arbitrary transmitter receiver distances are termed as large scale propagation
models. These are termed so because they predict the average signal strength for
large Tx-Rx separations, typically for hundreds of kilometers.

35
ld
or
Figure 4.1: Free space propagation model, showing the near and far fields.

Free Space Propagation Model

W
Although EM signals when traveling through wireless channels experience fading
effects due to various effects, but in some cases the transmission is with a direct
line of sight such as in satellite communication. Free space model predicts that
the received power decays as negative square root of the distance. Friis free space
TU
equation is given by
Pttttttrλ2
Pr (d) = (4.1)
(4π)2d2L
where Pt is the transmitted power, Pr(d) is the received power, ttt is the transmitter
antenna gain, ttr is the receiver antenna gain, d is the Tx-Rx separation and L is the
system loss factor depended upon line attenuation, filter losses and antenna losses
JN

and not related to propagation. The gain of the antenna is related to the effective
aperture of the antenna which in turn is dependent upon the physical size of the
antenna as given below
tt = 4πAe /λ2. (4.2)

The path loss, representing the attenuation suffered by the signal as it travels
ll

through the wireless channel is given by the difference of the transmitted and re-
ceived power in dB and is expressed as:
A

PL(dB) = 10 log Pt /Pr . (4.3)

36
The fields of an antenna can broadly be classified in two regions, the far field and
the near field. It is in the far field that the propagating waves act as plane waves
and the power decays inversely with distance. The far field region is also termed
as Fraunhofer region and the Friis equation holds in this region. Hence, the Friis
equation is used only beyond the far field distance, df , which is dependent upon the
largest dimension of the antenna as

ld
df = 2D2/λ. (4.4)

Also we can see that the Friis equation is not defined for d=0. For this reason, we

or
use a close in distance, do, as a reference point. The power received, Pr(d), is then
given by:
Pr (d) = Pr (do)(do/d)2. (4.5)
Ex. 1: Find the far field distance for a circular antenna with maximum dimension

λ=
W
of 1 m and operating frequency of 900 MHz.
Solution: Since the operating frequency f = 900 Mhz, the wavelength
3 × 108m/s m
900 10 6 Hz
×
TU
. Thus, with the largest dimension of the antenna, D=1m, the far field distance is
2D2 2(1)2
df = = = 6m
λ 0.33

.
Ex. 2: A unit gain antenna with a maximum dimension of 1 m produces 50 W
JN

power at 900 MHz. Find (i) the transmit power in dBm and dB, (ii) the received
power at a free space distance of 5 m and 100 m.
Solution:
(i) Tx power = 10log(50) = 17 dB = (17+30) dBm = 47 dBm
2
(ii) df = 2×D2 = 2 ×1 = 6m
λ 1/3
ll

Thus the received power at 5 m can not be calculated using free space distance
formula.
A

At 100 m ,
PT ttT ttRλ2
PR =
4πd2
50 × 1 × (1/3)2
=
4π1002

37
= 3.5 × 10−3 mW
PR (dBm) = 10logPr(mW ) = −24.5dBm

Basic Methods of Propagation

Reflection, diffraction and scattering are the three fundamental phenomena that

ld
cause signal propagation in a mobile communication system, apart from LoS com-
munication. The most important parameter, predicted by propagation models based
on above three phenomena, is the received power. The physics of the above phe-
nomena may also be used to describe small scale fading and multipath propagation.

or
The following subsections give an outline of these phenomena.

Reflection

W
Reflection occurs when an electromagnetic wave falls on an object, which has very
large dimensions as compared to the wavelength of the propagating wave. For ex-
ample, such objects can be the earth, buildings and walls. When a radio wave falls
on another medium having different electrical properties, a part of it is transmitted
into it, while some energy is reflected back. Let us see some special cases. If the
TU

medium on which the e.m. wave is incident is a dielectric, some energy is reflected
back and some energy is transmitted. If the medium is a perfect conductor, all
energy is reflected back to the first medium. The amount of energy that is reflected
back depends on the polarization of the e.m. wave.
JN

Another particular case of interest arises in parallel polarization, when no re-


flection occurs in the medium of origin. This would occur, when the incident angle
would be such that the reflection coefficient is equal to zero. This angle is the
Brewster’s angle. By applying laws of electro-magnetics, it is found to be
.
s1
sin(θB ) = . (4.6)
s1 + s2
ll

Further, considering perfect conductors, the electric field inside the conductor is
always zero. Hence all energy is reflected back. Boundary conditions require that
A

θi = θr (4.7)

and
Ei = Er (4.8)

57
for vertical polarization, and
Ei = −Er (4.9)

for horizontal polarization.

Diffraction

ld
Diffraction is the phenomenon due to which an EM wave can propagate beyond the
horizon, around the curved earth’s surface and obstructions like tall buildings. As
the user moves deeper into the shadowed region, the received field strength decreases.
But the diffraction field still exists an it has enough strength to yield a good signal.

or
This phenomenon can be explained by the Huygen’s principle, according to which,
every point on a wavefront acts as point sources for the production of sec- ondary
wavelets, and they combine to produce a new wavefront in the direction of

W
propagation. The propagation of secondary wavelets in the shadowed region results
in diffraction. The field in the shadowed region is the vector sum of the electric field
components of all the secondary wavelets that are received by the receiver.

Scattering
TU

The actual received power at the receiver is somewhat stronger than claimed by the
models of reflection and diffraction. The cause is that the trees, buildings and lamp-
posts scatter energy in all directions. This provides extra energy at the receiver.
Roughness is tested by a Rayleigh criterion, which defines a critical height hc of
JN

surface protuberances for a given angle of incidence θi, given by,


λ
hc = . (4.10)
8sinθi
A surface is smooth if its minimum to maximum protuberance h is less than hc ,
and rough if protuberance is greater than hc . In case of rough surfaces, the surface
reflection coefficient needs to be multiplied by a scattering loss factor ρS , given by
ll

πσhsinθ i 2
ρS = exp(−8( ) ) (4.11)
λ
A

where σh is the standard deviation of the Gaussian random variable h. The following
result is a better approximation to the observed value
πσ hsinθ i 2 πσhsinθ i 2
ρS = exp(−8( ) )I0[−8( ) ] (4.12)
λ λ

58
ld
or
W
Figure 4.2: Two-ray reflection model.

which agrees very well for large walls made of limestone. The equivalent reflection
TU
coefficient is given by,
Γrough = ρS Γ. (4.13)

Two Ray Reflection Model


JN

Interaction of EM waves with materials having different electrical properties than


the material through which the wave is traveling leads to transmitting of energy
through the medium and reflection of energy back in the medium of propagation.
The amount of energy reflected to the amount of energy incidented is represented
by Fresnel reflection coefficient Γ, which depends upon the wave polarization, angle
ll

of incidence and frequency of the wave. For example, as the EM waves can not pass
through conductors, all the energy is reflected back with angle of incidence equal to
the angle of reflection and reflection coefficient Γ = −1. In general, for parallel and
A

perpendicular polarizations, Γ is given by:

Γ|| = Er /Ei = η2 sin θt − η1 sin θ i/η2 sin θ t + η1 sin θ i (4.14)

59
Γ⊥ = Er/Ei = η2 sin θi − η1 sin θt/η2 sin θi + η1 sin θt. (4.15)

Seldom in communication systems we encounter channels with only LOS paths and
hence the Friis formula is not a very accurate description of the communication link.
A two-ray model, which consists of two overlapping waves at the receiver, one direct
path and one reflected wave from the ground gives a more accurate description as

ld
shown in Figure 4.2. A simple addition of a single reflected wave shows that power
varies inversely with the forth power of the distance between the Tx and the Rx.
This is deduced via the following treatment. From Figure 4.2, the total transmitted
and received electric fields are

or
ETT OT = Ei + ELOS , (4.16)

ETROT = Eg + ELOS . (4.17)

where
W
Let E0 is the free space electric field (in V/m) at a reference distance d0. Then
E d
E(d, t) = 0 0 cos(ω t − φ)
d

d
c
(4.18)

φ = ωc (4.19)
TU
c
and d > d0. The envelop of the electric field at d meters from the transmitter at
any time t is therefore
E0d0
|E(d, t)| = . (4.20)
d
This means the envelop is constant with respect to time.
JN

Two propagating waves arrive at the receiver, one LOS wave which travels a
t tt
distance of d and another ground reflected wave, that travels d . Mathematically,
it can be expressed as:
t E0 d0 t
E(d , t) = cos(ω t − φ ) (4.21)
c
dt
where
ll

t t
φ = ωc d (4.22)
c
and
E0 d0
A

tt tt
E(d , t) = cos(ω t − φ ) (4.23)
tt c
d
where
tt
tt d
φ = ωc . (4.24)
c

60
ld
Figure 4.3: Phasor diagram of electric Figure 4.4: Equivalent phasor diagram of
Figure 4.3.

or
fields.

According to the law of reflection in a dielectric, θi = θ 0 and Eg = ΓEi which means


the total electric field,
Et = Ei + Eg = Ei (1 + Γ). (4.25)

i.e.,
W
For small values of θ i, reflected wave is equal in magnitude and 180o out of phase
with respect to incident wave. Assuming perfect horizontal electric field polarization,

Γ⊥ = −1 =⇒ Et = (1 − 1)Ei = 0, (4.26)
TU

the resultant electric field is the vector sum of ELOS and Eg . This implies that,

ETROT = |ELOS + Eg |. (4.27)

It can be therefore written that


JN

ET OT E0d0 t E0d0 tt

R (d, t) = cos(ωc t − φ ) + (−1) cos(ωc t − φ ) (4.28)


dt dtt
In such cases, the path difference is
tt t
. .
∆ =d −d = (h t + h r )2 + d2 − (h t − h r)2 + d2 . (4.29)

However, when T-R separation distance is very large compared to (ht + hr), then
ll

2hthr
∆ ≈ (4.30)
d
A

Ex 3: Prove the above two equations, i.e., equation (4.29) and (4.30).
Once the path difference is known, the phase difference is
∆ωc
θ∆ = 2π∆ = (4.31)
λ λ

61
and the time difference,
∆ θ∆
τd = = . (4.32)
c 2πfc
When d is very large, then ∆ becomes very small and therefore ELOS and Eg are
virtually identical with only phase difference,i.e.,
E0d0 E0d0 E0d0
| |≈| | ≈| |. (4.33)

ld
t
d d dtt
tt
Say, we want to evaluate the received E-field at any t = dc . Then,
t
tt E0d0 tt d E 0 d0 tt tt
ET OT d d d d
(d, t = ) = dt cos(ω c − ωc c ) − tt cos(ωc − ωc )
c (4.34)

or
R
c E d ∆ωc d c c
= 0 0 cos( ) E0d0
t − cos(0o) (4.35)
d c dtt
E0d0
E 0 d0 ƒ
= θ∆ − (4.36)
dt dtt

we get

|ET OT
R (d)| =
.
( +
E0d0
≈ d (ƒ θ∆ − 1).

E0d0 E0d0
cos(θ ∆
W
Using phasor diagram concept for vector addition as shown in Figures 4.3 and 4.4,

2
)) + (
E0d0
sin(θ ∆))
2
(4.38)
(4.37)

d d d
TU

E0 d 0 . 2 2
= (cos(θ ) − 1) + sin (θ ∆ ) (4.39)

d
E0 d 0 ,
=
2 − 2cosθ (4.40)

d
E0 d0 θ
=2 sin( ∆ ). (4.41)
d 2
JN

For θ∆ < 0.5rad, sin( θ∆ ) ≈ θ∆ . Using equation (4.31) and further equation (4.30),
2 2 2
we can then approximate that
θ∆ π
sin( ) ≈ ∆ = 2πht hr < 0.5rad. (4.42)
2 λ λd

This raises the wonderful concept of ‘cross-over distance’ dc, defined as


ll

4πhthr
d > dc = 20πhthr = . (4.43)
5λ λ
A

The corresponding approximate received electric field is

ET OT E0d0 2πhthr hthr


R (d) ≈ 2 =k . (4.44)
d λd d2

62
Therefore, using equation (4.43) in (4.1), we get the received power as
Pttttttr h2th2r
Pr =
. (4.45)
Ld4
The cross-over distance shows an approximation of the distance after which the
received power decays with its fourth order. The basic difference between equation
(4.1) and (4.45) is that when d < dc , equation (4.1) is sufficient to calculate the

ld
path loss since the two-ray model does not give a good result for a short distance
due to the oscillation caused by the constructive and destructive combination of the
two rays, but whenever we distance crosses the ‘cross-over distance’, the power falls

or
off rapidly as well as two-ray model approximation gives better result than Friis
equation.
Observations on Equation (4.45): The important observations from this
equation are:

cross-over distance. W
1. This equation gives fair results when the T-R separation distance crosses the

1. In that case, the power decays as the fourth power of distance


K
TU
Pr(d) = , (4.46)
d4
with K being a constant.
2. Path loss is independent of frequency (wavelength).
3. Received power is also proportional to h2 and h2, meaning, if height of any of the
t r

antennas is increased, received power increases.


JN

Diffraction

Diffraction is the phenomena that explains the digression of a wave from a straight
line path, under the influence of an obstacle, so as to propagate behind the obstacle.
ll

It is an inherent feature of a wave be it longitudinal or transverse. For e.g the


sound can be heard in a room, where the source of the sound is another room
A

without having any line of sight. The similar phenomena occurs for light also but
the diffracted light intensity is not noticeable. This is because the obstacle or slit
need to be of the order of the wavelength of the wave to have a significant effect.
Thus radiation from a point source radiating in all directions can be received at any

63
ld
or
Figure 4.5: Huygen’s secondary wavelets.

W
point, even behind an obstacle (unless it is not completely enveloped by it), as shown
in Figure 4.5. Though the intensity received gets smaller as receiver is moved into the
shadowed region. Diffraction is explained by Huygens-Fresnel principle which states
that all points on a wavefront can be considered as the point source for secondary
wavelets which form the secondary wavefront in the direction of the prorogation.
TU
Normally, in absence of an obstacle, the sum of all wave sources is zero at a point
not in the direct path of the wave and thus the wave travels in the straight line. But
in the case of an obstacle, the effect of wave source behind the obstacle cannot be
felt and the sources around the obstacle contribute to the secondary wavelets in the
JN

shadowed region, leading to bending of wave. In mobile communication, this has a


great advantage since, by diffraction (and scattering, reflection), the receiver is able
to receive the signal even when not in line of sight of the transmitter. This we show
in the subsection given below.

Knife-Edge Diffraction Geometry


ll

As shown in Figure 4.6, consider that there’s an impenetrable obstruction of hight


h at a distance of d1 from the transmitter and d2 from the receiver. The path
A

difference between direct path and the diffracted path is


. .
δ= 2 + d2 + h2 − (d1 + d 2)
d21 + h 2 (4.47)

64
ld
or
Figure 4.6: Diffraction through a sharp edge.

which can be further simplified as

1
W
δ = d1 (1 + h2 /2d2 ) + d2 (1 + h2 /2d2 ) − (d1 + d2 )
2

= h2 /(2d1 ) + h2 /(2d2 ) = h2 (d1 + d2 )/(2d1 d2 ). (4.48)


TU
Thus the phase difference equals

φ = 2πδ/λ = 2πh 2(d 1 + d 2)/λ2(d1 d 2). (4.49)

With the following considerations that


JN

α=β +γ (4.50)

and
α ≈ tanα (4.51)

we can write,
ll

αtanα = tanβ + tanγ = h/d1 + h/d2 = h(d1 + d 2)/d 1 d 2. (4.52)

In order to normalize this, we usually use a Fresnel-Kirchoff diffraction parameter


A

v, expressed as
. .
v = h 2(d1 + d2 )/(λd1 d2 ) = α (2d1 d2 )/(λ(d1 + d2 )) (4.53)

65
ld
or
W
TU

Figure 4.7: Fresnel zones.

and therefore the phase difference becomes

φ = πv 2 /2. (4.54)
JN

From this, we can observe that: (i) phase difference is a function of the height of
the obstruction, and also, (ii) phase difference is a function of the position of the
obstruction from transmitter and receiver.

Fresnel Zones: the Concept of Diffraction Loss


ll

As mentioned before, the more is the object in the shadowed region greater is the
diffraction loss of the signal. The effect of diffraction loss is explained by Fresnel
A

zones as a function of the path difference. The successive Fresnel zones are limited
by the circular periphery through which the path difference of the secondary waves
is nλ/2 greater than total length of the LOS path, as shown in Figure 4.7. Thus
successive Fresnel zones have phase difference of π which means they alternatively

66
provide constructive and destructive interference to the received the signal. The
radius of the each Fresnel zone is maximum at middle of transmitter and receiver

(i.e. when d1 = d2 ) and decreases as moved to either side. It is seen that the loci
of a Fresnel zone varied over d1 and d2 forms an ellipsoid with the transmitter and
receiver at its focii. Now, if there’s no obstruction, then all Fresnel zones result in
only the direct LOS prorogation and no diffraction effects are observed. But if an

ld
obstruction is present, depending on its geometry, it obstructs contribution from
some of the secondary wavelets, resulting in diffraction and also the loss of energy,
which is the vector sum of energy from unobstructed sources. please note that height

or
of the obstruction can be positive zero and negative also. The diffraction losses are
minimum as long as obstruction doesn’t block volume of the 1st Fresnel zone. As a
rule of thumb, diffraction effects are negligible beyond 55% of 1st Fresnel zone.
Ex 4: Calculate the first Fresnel zone obstruction height maximum for f = 800
MHz.
Solution:

λ=
c
f
=
W 3
×
108 3
8 × 102 × 106 = 8 m
.
TU
λ(d1+d2 )
H= . d1+d2
3
250×250
H1 = 8
= 6.89m
500
Thus H1 = 10 + 6.89 = 16.89m
JN

(b)
. 100 × 400 .
3 ×
H2 = 8 = 10 (0.3) = 5.48m
500
Thus

H2 = 10 + 5.6 = 15.48m

. To have good power strength, obstacle should be within the 60% of the first fresnel
ll

zone.
A

Ex 5: Given f=900 MHz, d1 = d2 = 1 km, h = 25m, where symbols have usual


meaning. Compute the diffraction loss. Also find out in which Fresnel zone the tip
of the obstruction lies.

67
ld
or
Figure 4.8: Knife-edge Diffraction Model

Given,
W
ttd(dB) = 20 log(0.5 − 0.62v)

ttd (dB) = 20 log(0.225/v)


− 1 < v <= 0

v > 2.24
TU

Solution: . .
2(d1 + d 2) 2 × 2000
v=h = 2.74
λd1d2 = 25 1
3 10
Gd(dB) = 20 log( 225 ) = −21.7dB
v
JN

Since loss = -ttd (dB) = 21.7 dB

(2.74)2
n= = 3.5
2
ll

Thus n=4.

4.5.3 Knife-edge diffraction model


A

Knife-edge diffraction model is one of the simplest diffraction model to estimate the
diffraction loss. It considers the object like hill or mountain as a knife edge sharp

68
object. The electric field strength, Ed of a knife-edge diffracted wave is given by
¸ ∞
Ed/Eo = F (v) = (1 + j)/2 (exp((−jπt2 )/2)dt. (4.55)
v

The diffraction gain due to presence of knife edge can be given as

ttd(db) = 20log|F (v)| (4.56)

ld
ttd (db) = 0v <= −1 (4.57)

ttd(db) = 20log(0.5 − 0.62) − 1 <= v <= 0 (4.58)

or
ttd(db) = 20log(0.5exp(−0.95v)) 0 <= v <= 1 (4.59)

ttd(db) = 20log(0.4 − sqrt(0.1184 − (0.38 − 0.1v2))) 1 <= v <= 2.4 (4.60)

ttd (db) = 20log(0.225/v) v > 2.4 (4.61)

W
When there are more than one obstruction, then the equivalent model can be found
by one knife-edge diffraction model as shown in Figure 4.8.

Link Budget Analysis


TU

Log-distance Path Loss Model

According to this model the received power at distance d is given by,


d d
PL(d)( )n =⇒ PL(dB) = PL(d ) + 10nlog( ) (4.62)
0
d0 d0
JN

The value of n varies with propagation environments. The value of n is 2 for free
space. The value of n varies from 4 to 6 for obstruction of building, and 3 to 5 for

urban scenarios. The important factor is to select the correct reference distance d0.
For large cell area it is 1 Km, while for micro-cell system it varies from 10m-1m.
Limitations:
ll

Surrounding environmental clutter may be different for two locations having


the same transmitter to receiver separation. Moreover it does not account for the
shadowing effects.
A

69
Log Normal Shadowing

The equation for the log normal shadowing is given by,


d
PL(dB) = PL(dB) + Xσ = PL(d0) + 10nlog( ) + Xσ (4.63)
d0

where Xσ is a zero mean Gaussian distributed random variable in dB with standard

ld
deviation σ also in dB. In practice n and σ values are computed from measured
data.

Average received power

or
The ‘Q’ function is given by,
z
Q(z) = 0.5(1 − erf ( √ )) (4.64)
2
and

γ is
W
Q(z) = 1 − Q(−z)

γ − Pr
(4.65)

So the probability that the received signal level (in dB) will exceed a certain value
TU
P (Pd > γ) = Q( ). (4.66)
σ

Outdoor Propagation Models

There are many empirical outdoor propagation models such as Longley-Rice model,
JN

Durkin’s model, Okumura model, Hata model etc. Longley-Rice model is the most
commonly used model within a frequency band of 40 MHz to 100 GHz over different
terrains. Certain modifications over the rudimentary model like an extra urban
factor (UF) due to urban clutter near the reciever is also included in this model.
Below, we discuss some of the outdoor models, followed by a few indoor models too.
ll

Okumura Model

The Okumura model is used for Urban Areas is a Radio propagation model that is
A

used for signal prediction.The frequency coverage of this model is in the range of
200 MHz to 1900 MHz and distances of 1 Km to 100 Km.It can be applicable for
base station effective antenna heights (ht) ranging from 30 m to 1000 m.

70
Okumura used extensive measurements of base station-to-mobile signal attenua-
tion throughout Tokyo to develop a set of curves giving median attenuation relative
to free space (A mu) of signal propogation in irregular terrain. The empirical path-
loss formula of Okumura at distance d parameterized by the carrier frequency fc is
given by

ld
PL(d)dB = L(fc , d) + Amu (fc , d) − tt(ht) − tt(hr) − ttAREA (4.67)

where L(fc, d) is free space path loss at distance d and carrier frequency fc , A mu(fc , d)
is the median attenuation in addition to free-space path loss across all environments,tt(ht)

or
is the base station antenna height gain factor,tt(hr) is the mobile antenna height gain
factor,ttAREA is the gain due to type of environment. The values of Amu(fc, d) and
ttAREA are obtained from Okumura’s empirical plots. Okumura derived empirical
formulas for tt(ht) and tt(hr) as follows:

W
tt(ht ) = 20 log10 (ht /200), 30m < ht < 1000m

tt(hr ) = 10 log10 (hr /3),

tt(hr ) = 20 log10 (hr /3),


hr ≤ 3m

3m < hr < 10m


(4.68)

(4.69)

(4.70)
TU

Correlation factors related to terrain are also developed in order to improve the
models accuracy. Okumura’s model has a 10-14 dB empirical standard deviation
between the path loss predicted by the model and the path loss associated with one
of the measurements used to develop the model.
JN

Hata Model

The Hata model is an empirical formulation of the graphical path-loss data provided
by the Okumura and is valid over roughly the same range of frequencies, 150-1500
MHz. This empirical formula simplifies the calculation of path loss because it is
closed form formula and it is not based on empirical curves for the different param-
ll

eters. The standard formula for empirical path loss in urban areas under the Hata
model is
A

PL,urban (d)dB = 69.55+26.16 log10 (fc )−13.82 log10 (ht )−a(hr )+(44.9−6.55 log10 (ht )) log10 (d)
(4.71)

71
The parameters in this model are same as in the Okumura model,and a(hr) is a
correction factor for the mobile antenna height based on the size of coverage area.For
small to medium sized cities this factor is given by

a(hr ) = (1.11 log10(fc ) − 0.7)hr − (1.56 log10(fc ) − 0.8)dB

and for larger cities at a frequencies fc > 300 MHz by

ld
a(hr) = 3.2(log10(11.75hr))2 − 4.97dB

else it is
a(hr) = 8.29(log10(1.54hr))2 − 1.1dB

or
Corrections to the urban model are made for the suburban, and is given by

PL,suburban (d)dB = PL,urban (d)dB − 2(log10 (fc /28))2 − 5.4 (4.72)

Unlike the Okumura model,the Hata model does not provide for any specific path-

W
correlation factors. The Hata model well approximates the Okumura model for
distances d > 1 Km. Hence it is a good model for first generation cellular systems,
but it does not model propogation well in current cellular systems with smaller cell
sizes and higher frequencies. Indoor environments are also not captured by the Hata
TU
model.

Indoor Propagation Models

The indoor radio channel differs from the traditional mobile radio channel in ways
JN

- the distances covered are much smaller ,and the variability of the environment
is much greater for smaller range of Tx-Rx separation distances.Features such as lay-
out of the building,the construction materials,and the building type strongly in-
fluence the propagation within the building.Indoor radio propagation is dominated
by the same mechanisms as outdoor: reflection, diffraction and scattering with vari-
able conditions. In general,indoor channels may be classified as either line-of-sight
ll

or obstructed.
A

Partition Losses Inside a Floor (Intra-floor)

The internal and external structure of a building formed by partitions and obstacles
vary widely.Partitions that are formed as a part of building structure are called

72
hard partitions , and partitions that may be moved and which do not span to
the ceiling are called soft partitions. Partitions vary widely in their physical and
electrical characteristics,making it difficult to apply general models to specific indoor
installations.

Partition Losses Between Floors (Inter-floor)

ld
The losses between floors of a building are determined by the external dimensions
and materials of the building,as well as the type of construction used to create the
floors and the external surroundings. Even the number of windows in a building

or
and the presence of tinting can impact the loss between floors.

Log-distance Path Loss Model

W
It has been observed that indoor path loss obeys the distance power law given by

PL(dB) = PL(d 0 ) + 10n log10(d/d0) + Xσ

where n depends on the building and surrounding type, and Xσ represents a normal
random variable in dB having standard deviation of σ dB.
(4.73)
TU

Summary

In this chapter, three principal propagation models have been identified: free-space
propagation, reflection and diffraction, which are common terrestrial models and
JN

these mainly explains the large scale path loss. Regarding path-loss, one important
factor introduced in this chapter is log-distance path loss model. These, however,
may be insignificant when we consider the small-scale rapid path losses. This is
discussed in the next chapter.
ll

References

1. T. S. Rappaport, Wireless Communications: Principles and Practice, 2nd ed.


A

Singapore: Pearson Education, Inc., 2002.

2. S. Haykin and M. Moher, Modern Wireless Communications. Singapore: Pear-


son Education, Inc., 2002.

73
3. J. W. Mark and W. Zhuang, Wireless Communications and Networking. New
Delhi: PHI, 2005.

ld
or
W
TU
JN
ll
A

74
Chapter 5

ld
Multipath Wave Propagation

or
and Fading

Multipath Propagation
W
In wireless telecommunications, multipath is the propagation phenomenon that re-
sults in radio signals reaching the receiving antenna by two or more paths. Causes
of multipath include atmospheric ducting, ionospheric reflection and refraction, and
TU

reflection from water bodies and terrestrial objects such as mountains and buildings.
The effects of multipath include constructive and destructive interference, and phase
shifting of the signal. In digital radio communications (such as GSM) multipath can
cause errors and affect the quality of communications. We discuss all the related
JN

issues in this chapter.

Multipath & Small-Scale Fading

Multipath signals are received in a terrestrial environment, i.e., where different forms
of propagation are present and the signals arrive at the receiver from transmitter via
ll

a variety of paths. Therefore there would be multipath interference, causing multi-


path fading. Adding the effect of movement of either Tx or Rx or the surrounding
A

clutter to it, the received overall signal amplitude or phase changes over a small
amount of time. Mainly this causes the fading.

75
Fading

The term fading, or, small-scale fading, means rapid fluctuations of the amplitudes,
phases, or multipath delays of a radio signal over a short period or short travel
distance. This might be so severe that large scale radio propagation loss effects
might be ignored.

ld
Multipath Fading Effects

In principle, the following are the main multipath effects:

or
1. Rapid changes in signal strength over a small travel distance or time interval.

2. Random frequency modulation due to varying Doppler shifts on different mul-


tipath signals.

Factors Influencing FadingW


3. Time dispersion or echoes caused by multipath propagation delays.

The following physical factors influence small-scale fading in the radio propagation
channel:
TU

(1) Multipath propagation – Multipath is the propagation phenomenon that re-


sults in radio signals reaching the receiving antenna by two or more paths.
The effects of multipath include constructive and destructive interference, and
phase shifting of the signal.
JN

(2) Speed of the mobile – The relative motion between the base station and the
mobile results in random frequency modulation due to different doppler shifts
on each of the multipath components.

(3) Speed of surrounding objects – If objects in the radio channel are in mo-
tion, they induce a time varying Doppler shift on multipath components. If
ll

the surrounding objects move at a greater rate than the mobile, then this effect
dominates fading.
A

(4) Transmission Bandwidth of the signal – If the transmitted radio signal


bandwidth is greater than the “bandwidth” of the multipath channel (quanti-
fied by coherence bandwidth), the received signal will be distorted.

76
Types of Small-Scale Fading

The type of fading experienced by the signal through a mobile channel depends
on the relation between the signal parameters (bandwidth, symbol period) and the
channel parameters (rms delay spread and Doppler spread). Hence we have four
different types of fading. There are two types of fading due to the time dispersive

ld
nature of the channel.

Fading Effects due to Multipath Time Delay Spread

or
Flat Fading

Such types of fading occurs when the bandwidth of the transmitted signal is less than
the coherence bandwidth of the channel. Equivalently if the symbol period of the
signal is more than the rms delay spread of the channel, then the fading is flat fading.

W
So we can say that flat fading occurs when

BS BC (5.1)
TU
where BS is the signal bandwidth and BC is the coherence bandwidth. Also

TS στ (5.2)

where TS is the symbol period and στ is the rms delay spread. And in such a case,
mobile channel has a constant gain and linear phase response over its bandwidth.
JN

Frequency Selective Fading

Frequency selective fading occurs when the signal bandwidth is more than the co-
herence bandwidth of the mobile radio channel or equivalently the symbols duration
of the signal is less than the rms delay spread.
ll

BS BC (5.3)
A

and
TS στ (5.4)

77
At the receiver, we obtain multiple copies of the transmitted signal, all attenuated
and delayed in time. The channel introduces inter symbol interference. A rule of
thumb for a channel to have flat fading is if
στ
0.1 (5.5)

TS

ld
Fading Effects due to Doppler Spread

Fast Fading

In a fast fading channel, the channel impulse response changes rapidly within the

or
symbol duration of the signal. Due to Doppler spreading, signal undergoes frequency
dispersion leading to distortion. Therefore a signal undergoes fast fading if

TS TC (5.6)

where TC is the coherence time and

W
BS BD (5.7)

where BD is the Doppler spread. Transmission involving very low data rates suffer
TU
from fast fading.

Slow Fading

In such a channel, the rate of the change of the channel impulse response is much
less than the transmitted signal. We can consider a slow faded channel a channel in
JN

which channel is almost constant over atleast one symbol duration. Hence

TS TC (5.8)

and
BS BD (5.9)
ll

We observe that the velocity of the user plays an important role in deciding whether
the signal experiences fast or slow fading.
A

78
ld
or
Figure 5.1: Illustration of Doppler effect.

Doppler Shift

W
The Doppler effect (or Doppler shift) is the change in frequency of a wave for an
observer moving relative to the source of the wave. In classical physics (waves in
a medium), the relationship between the observed frequency f and the emitted
TU
frequency fo is given by:
. .
v ± vr
f = f0 (5.10)
v ± vs
where v is the velocity of waves in the medium, vs is the velocity of the source
relative to the medium and vr is the velocity of the receiver relative to the medium.
In mobile communication, the above equation can be slightly changed according
JN

to our convenience since the source (BS) is fixed and located at a remote elevated
level from ground. The expected Doppler shift of the EM wave then comes out to
vr vr
be ± fo or, ± . As the BS is located at an elevated place, a cos φ factor would
c λ

also be multiplied with this. The exact scenario, as given in Figure 5.1, is illustrated
below.
ll

Consider a mobile moving at a constant velocity v, along a path segment length


d between points A and B, while it receives signals from a remote BS source S. The
A

difference in path lengths traveled by the wave from source S to the mobile at points
A and B is ∆l = d cos θ = v∆t cos θ, where ∆t is the time required for the mobile
to travel from A to B, and θ is assumed to be the same at points A and B since the

79
source is assumed to be very far away. The phase change in the received signal due
to the difference in path lengths is therefore

2π∆l 2πv∆t
∆ϕ = cos θ (5.11)
λ = λ

and hence the apparent change in frequency, or Doppler shift (fd) is


1 ∆ϕ v

ld
fd = . = . cos θ. (5.12)
2π ∆t λ
Example 1
An aircraft is heading towards a control tower with 500 kmph, at an elevation of 20◦ .

or
Communication between aircraft and control tower occurs at 900 MHz. Find out
the expected Doppler shift.
Solution As given here,
v = 500 kmph

W
the horizontal component of the velocity is

vt = v cos θ = 500 × cos 20◦ = 130 m/s

Hence, it can be written that


TU
900 × 106 1
λ= = m
3 × 108 3
130
fd = = 390 Hz
1/3

If the plane banks suddenly and heads for other direction, the Doppler shift change
JN

will be 390 Hz to −390 Hz.

Impulse Response Model of a Multipath Channel

Mobile radio channel may be modeled as a linear filter with time varying impulse
response in continuous time. To show this, consider time variation due to receiver
ll

motion and time varying impulse response h(d, t) and x(t), the transmitted signal.
The received signal y(d, t) at any position d would be
¸
A


y(d, t) = x(t) ∗ h(d, t) = x(τ ) h(d, t − τ ) dτ (5.13)
−∞
¸ ∞
For a causal system: h(d, t) = 0, for t < 0 and for a stable system |h(d, t)| dt <
−∞

80
Applying causality condition in the above equation, h(d, t − τ ) = 0 for t − τ < 0
⇒ τ > t, i.e., the integral limits are changed to

¸ t
y(d, t) = x(τ ) h(d, t − τ ) dτ.
−∞

ld
Since the receiver moves along the ground at a constant velocity v, the position of
the receiver is d = vt, i.e.,
¸ t
y(vt, t) = x(τ ) h(vt, t − τ ) dτ.
−∞

or
Since v is a constant, y(vt, t) is just a function of t. Therefore the above equation
can be expressed as
¸ t
y(t) = x(τ ) h(vt, t − τ ) dτ = x(t) ∗ h(vt, t) = x(t) ∗ h(d, t) (5.14)
−∞

W
It is useful to discretize the multipath delay axis τ of the impulse response into equal
time delay segments called excess delay bins, each bin having a time delay width

equal to ( τi+1 − τi ) = ∆τ and τi = i∆τ for i ∈ {0, 1, 2, ..N − 1}, where N represents
the total number of possible equally-spaced multipath components, including the
TU

first arriving component. The useful frequency span of the model is 2/∆τ . The
model may be used to analyze transmitted RF signals having bandwidth less than
2/ .
∆τ
If there are N multipaths, maximum excess delay is given by N ∆τ .
JN

{y(t) = x(t) ∗ h(t, τi )|i = 0, 1, ...N − 1} (5.15)

Bandpass channel impulse response model is

x(t) → h(t, τ ) = Re{hb (t, τ )ejωc t → y(t) = Re{r(t)ejωc t } (5.16)

Baseband equivalent channel impulse response model is given by


ll

1 1
c(t) → hb(t, τ ) → r(t) = c(t) ∗ hb(t, τ ) (5.17)
2 2
A

Average power is 1
x2(t) = c(t) 2 (5.18)
| |
2

81
The baseband impulse response of a multipath channel can be expressed as
N−1
.
hb(t, τ ) = ai(t, τ ) exp[j(2πfcτi(t) + ϕi(t, τ ))]δ(τ − τi(t)) (5.19)
i=0

where ai(t, τ ) and τi(t) are the real amplitudes and excess delays, respectively, of
the ith multipath component at time t. The phase term 2πfc τ i(t) + ϕi(t, τ ) in the
above equation represents the phase shift due to free space propagation of the ith

ld
multipath component, plus any additional phase shifts which are encountered in the
channel.
If the channel impulse response is wide sense stationary over a small-scale time or

or
distance interval, then
N−1
.
hb(τ ) = ai exp[jθi ]δ(τ − τi ) (5.20)
i=0

For measuring hb(τ ), we use a probing pulse to approximate δ(t) i.e.,

W
p(t) ≈ δ(t − τ )

Power delay profile is taken by spatial average of |hb(t, τ )|

received power delay profile in a local area is given by


2 over a local area. The
(5.21)
TU
p(τ ) ≈ k|hb (t; τ )| 2. (5.22)

Relation Between Bandwidth and Received Power

In actual wireless communications, impulse response of a multipath channel is mea-


JN

sured using channel sounding techniques. Let us consider two extreme channel
sounding cases.
Consider a pulsed, transmitted RF signal

x(t) = Re{p(t)ej2πfc t } (5.23)


.
4τmax
where p(t) = for 0 ≤ t ≤ Tbb and 0 elsewhere. The low pass channel output
ll

T bb

is
N−1
.
ai exp[jθi ]p(t − τi )
A

r(t) = 2
i=0 .
N−1 τmax Tb
.
= ai exp[jθi ]. − τi).
Tbb rect(t −
i=0 2

82
ld
or
Figure 5.2: A generic transmitted pulsed RF signal.

The received power at any time t0 is

|r(t0 )| 2 = 1
τmax
1
τmax
¸

0
τmax
¸
W
r(t)r∗ (t)dt
.N −1 .
1
TU
.
= a2k(t0)p2(t − τk) dt
τmax 4
0 k=0
τmax
1 .
N−1 ¸ .. .2
= a2 (t0) τ max rect(t − Tb − τ i) dt
k
τmax k=0 Tbb 2
0
N−1
.
= a2k(t0).
JN

k=0

Interpretation: If the transmitted signal is able to resolve the multipaths, then


average small-scale receiver power is simply sum of average powers received from
each multipath components.
N−1 N−1
ll

. .
2
Ea,θ [PW B ] = Ea,θ [ |ai exp(jθi )| ] ≈ a2i (5.24)
i=0 i=0

Now instead of a pulse, consider a CW signal, transmitted into the same channel
A

and for simplicity, let the envelope be c(t) = 2. Then


N−1
.
r(t) = ai exp[jθi(t, τ )] (5.25)
i=0

83
and the instantaneous power is
N−1
.
|r(t)|2 =| ai exp[jθi(t, τ )] |2 (5.26)
i=0

Over local areas, ai varies little but θ i varies greatly resulting in large fluctuations.
N−1
.
Ea,θ [PCW ] = Ea,θ [ |ai exp(jθi )| 2]

ld
i=0
N−1 N−1 N
. . .
≈ a2i + 2 rij cos(θi − θj )
i=0 i=0 i,jƒ=i

or
where rij = Ea [ai aj ].

If, rij = cos(θi − θj ) = 0, then Ea,θ [PCW ] = Ea,θ [PWB ]. This occurs if multipath

components are uncorrelated or if multipath phases are i.i.d over [0, 2π].
Bottomline:

W
1. If the signal bandwidth is greater than multipath channel bandwidth then
fading effects are negligible
TU
2. If the signal bandwidth is less than the multipath channel bandwidth, large
fading occurs due to phase shift of unresolved paths.

Linear Time Varying Channels (LTV)

The time variant transfer function(TF) of an LTV channel is FT of h(t, τ ) w.r.t. τ .


JN

¸ ∞
H(f, t) = F T [h(τ, t)] = h(τ, t)e−j 2πf τ dτ (5.27)
−∞
¸ ∞
h(τ, t) = F T −1 [H(f, t)] = H(f, t)ej2πf τ df (5.28)
−∞

The received signal


¸ ∞
r(t) = R(f, t)ej 2πf t df (5.29)
ll

−∞

where R(f, t) = H(f, t)X(f ).


.
A

For flat fading channel, h(τ, t) = Z(t)δ(τ − τi) where Z(t) = αn(t)e−j2πfcτn(t). In

this case, the received signal is


¸ ∞
r(t) = h(τ, t)x(t − τ ) dτ = Z(t)x(t − τi ) (5.30)
−∞

84
ld
or
where the channel becomes multiplicative.
Doppler spread functions:
W
Figure 5.3: Relationship among different channel functions.

¸
TU

H(f, ν) = F T [H(f, t)] = H(f, t)e−j 2πνt dt (5.31)
−∞

and ¸ ∞
H(f, t) = F T −1
[H(f, ν)] = H(f, ν)ej 2πνt dν (5.32)
−∞

Delay Doppler spread:


JN

¸ ∞
H(τ, ν) = F T [h(τ, t)] = h(τ, t)e−j 2πνt dt (5.33)
−∞

Small-Scale Multipath Measurements

Direct RF Pulse System


ll

A wideband pulsed bistatic radar usually transmits a repetitive pulse of width Tbb s,
2
and uses a receiver with a wide bandpass filter (BW = T bb
Hz). The signal is then
A

amplified, envelope detected, and displayed and stored on a high speed oscilloscope.
Immediate measurements of the square of the channel impulse response convolved
with the probing pulse can be taken. If the oscilloscope is set on averaging mode,
then this system provides a local average power delay profile.

85
ld
or
W
Figure 5.4: Direct RF pulsed channel IR measurement.
TU

This system is subject to interference noise. If the first arriving signal is blocked
or fades, severe fading occurs, and it is possible the system may not trigger properly.

Frequency Domain Channel Sounding


JN

In this case we measure the channel in the frequency domain and then convert it
into time domain impulse response by taking its inverse discrete Fourier transform
(IDFT). A vector network analyzer controls a swept frequency synthesizer. An S-
parameter test set is used to monitor the frequency response of the channel. The
sweeper scans a particular frequency band, centered on the carrier, by stepping
through discrete frequencies. The number and spacing of the frequency step impacts
ll

the time resolution of the impulse response measurement. For each frequency step,
the S-parameter test set transmits a known signal level at port 1 and monitors
A

the received signal at port 2. These signals allow the analyzer to measure the
complex response, S21(ω), of the channel over the measured frequency range. The
S21(ω) measure is the measure of the signal flow from transmitter antenna to receiver

86
ld
or
W
TU
Figure 5.5: Frequency domain channel IR measurement.

antenna (i.e., the channel).


This system is suitable only for indoor channel measurements. This system is
also non real-time. Hence, it is not suitable for time-varying channels unless the
JN

sweep times are fast enough.

Multipath Channel Parameters

To compare the different multipath channels and to quantify them, we define some
parameters. They all can be determined from the power delay profile. These pa-
ll

rameters can be broadly divided in to two types.


A

Time Dispersion Parameters

These parameters include the mean excess delay,rms delay spread and excess delay
spread. The mean excess delay is the first moment of the power delay profile and is

87
defined as

.
. τ P(τk)τk
τ¯ = .a2k k = . (5.34)
a2 k P(τk)
where ak is the amplitude, τk is the excess delay and P (τk) is the power of the
individual multipath signals.
The mean square excess delay spread is defined as

ld
.
P (τ )τ 2
τ¯2 = . k k (5.35)
P(τk )
Since the rms delay spread is the square root of the second central moment of the

or
power delay profile, it can be written as
.
στ = τ¯2 − (τ̄ )2 (5.36)

As a rule of thumb, for a channel to be flat fading the following condition must be
satisfied

TS W στ

0.1

where TS is the symbol duration. For this case, no equalizer is required at the
receiver.
(5.37)
TU
Example 2

1. Sketch the power delay profile and compute RMS delay spread for the follow-
ing:
1
.
P (τ ) = δ(τ − n × 10−6) (in watts)
n=0
JN

2. If BPSK modulation is used, what is the maximum bit rate that can be sent
through the channel without needing an equalizer?

Solution

1. P (0) = 1 watt, P (1) = 1 watt


ll

(1)(0) + (1)(1)
τ = = 0.5µs
1+1
A

τ 2 = 0.5µs2 στ = 0.5µs

88
ld
1
2. For flat fading channel, we need στ
0.1 ⇒ Rs = = 0.2 × 104 = 200 kbps
Ts Ts

For BPSK we need Rb = Rs = 200 kbps

Example 3 A simple delay spread bound: Feher’s upper bound

or
Consider a simple worst-case delay spread scenario as shown in figure below.

W
TU
Here dmin = d0 and dmax = di + dr
Transmitted power = PT , Minimum received power = PRmin = PT hreshold

PRmin λ
= ttT ttR ( )2
PT 4πdmax
JN

Put ttT = ttR = 1 i.e., considering omni-directional unity gain antennas

λ PT 1
dmax = ( )( )2
4π P Rmin
dmax λ PT 1
τmax = =( )( )2
c 4πc PRmin
1 PT 1
τmax =( )( )2
ll

4πf PRmin

Frequency Dispersion Parameters


A

To characterize the channel in the frequency domain, we have the following param-
eters.

89
(1) Coherence bandwidth: it is a statistical measure of the range of frequencies
over which the channel can be considered to pass all the frequency components with
almost equal gain and linear phase. When this condition is satisfied then we say the
channel to be flat.
Practically, coherence bandwidth is the minimum separation over which the two
frequency components are affected differently. If the coherence bandwidth is con-

ld
sidered to be the bandwidth over which the frequency correlation function is above
0.9, then it is approximated as
1
BC ≈ . (5.38)
50στ

or
However, if the coherence bandwidth is considered to be the bandwidth over which
the frequency correlation function is above 0.5, then it is defined as
1
. (5.39)
BC ≈

W 5στ
The coherence bandwidth describes the time dispersive nature of the channel in the
local area. A more convenient parameter to study the time variation of the channel
is the coherence time. This variation may be due to the relative motion between the
mobile and the base station or the motion of the objects in the channel.
TU

(2) Coherence time: this is a statistical measure of the time duration over which
the channel impulse response is almost invariant. When channel behaves like this,
it is said to be slow faded. Essentially it is the minimum time duration over which
two received signals are affected differently. For an example, if the coherence time
JN

is considered to be the bandwidth over which the time correlation is above 0.5, then
it can be approximated as
9
TC ≈ (5.40)
16πfm
where fm is the maximum doppler spread given be fm = νλ.
Another parameter is the Doppler spread (BD) which is the range of frequencies
ll

over which the received Doppler spectrum is non zero.

Statistical models for multipath propagation


A

Many multipath models have been proposed to explain the observed statistical na-
ture of a practical mobile channel. Both the first order and second order statistics

90
ld
or
Figure 5.6: Two ray NLoS multipath, resulting in Rayleigh fading.

W
have been examined in order to find out the effective way to model and combat the
channel effects. The most popular of these models are Rayleigh model, which de-
scribes the NLoS propagation. The Rayleigh model is used to model the statistical
time varying nature of the received envelope of a flat fading envelope. Below, we
discuss about the main first order and second order statistical models.
TU

NLoS Propagation: Rayleigh Fading Model

Let there be two multipath signals S1 and S2 received at two different time instants
due to the presence of obstacles as shown in Figure 5.6. Now there can either be
JN

constructive or destructive interference between the two signals.


Let En be the electric field and Θn be the relative phase of the various multipath
signals.So we have
N
.
Ẽ = En ej θn (5.41)
n=1

Now if N→ ∞(i.e. are sufficiently large number of multipaths) and all the En are
ll

IID distributed, then by Central Limit Theorem we have,


N
.
lim Ẽ = lim En ejθn (5.42)
A

N →∞
N →∞
n=1

= Zr + jZi = Rejφ (5.43)

91
where Zr and Zi are Gaussian Random variables. For the above case
.
2 (5.44)
R= Z 2r + Z i

and
Zi
φ = tan−1 (5.45)
Zr
For all practical purposes we assume that the relative phase Θn is uniformaly dis-

ld
tributed. 2π
jθn 1 ¸
E[e ]= ejθ dθ = 0 (5.46)

0

or
It can be seen that En and Θn are independent. So,
.
E[Ẽ] = E[ En ejθn ] = 0 (5.47)
. .2 . . .. .
N
E[. Ẽ . ] = E[ En ejθn E ∗ e−jθn ] = E[ En Em ej (θn −θ m ) ] = E 2 = P0
. . n

W m n

where P0 is the total power obtained. To find the Cumulative Distribution Func-
tion(CDF) of R, we proceed as follows.
¸ ¸
n=1
n

(5.48)

FR (r) = Pr (R ≤ r) = fZi,Zr (zi , zr)dzi dzr (5.49)


TU

where A is determined by the values taken by the dummy variable r. Let Zi and Zr
be zero mean Gaussian RVs. Hence the CDF can be written as
¸ ¸ −(Z2 +Z 2 )

FR(r) = √ 1 e r
2σ2
i
dZidZr (5.50)
2πσ2
JN

Let Zr = p cos(Θ) and Zi = p sin(Θ) So we have


2π 2π
¸ ¸ 1 −p 2
FR(r) = √ e 2σ2 pdpdθ (5.51)
2πσ 2
0 0
−r2

= 1 − e 2σ2 (5.52)
ll

Above equation is valid for all r ≥ 0. The pdf can be written as

− r2
A

r 2
(5.53)
fR(r) = e 2σ
σ2
and is shown in Figure 5.7 with different σ values. This equation too is valid for all
r ≥ 0. Above distribution is known as Rayleigh distribution and it has been derived

92
ld
or
Figure 5.7: Rayleigh probability density function.

for slow fading. However, if fD 1 Hz, we call it as Quasi-stationary Rayleigh


fading. We observe the following:

W
E[R] =
.

E[R2 ] = 2σ2

var[R] = (2
π
2

π 2
− 2

σ (5.54)

(5.55)

(5.56)
TU
median[R] = 1.77σ. (5.57)

LoS Propagation: Rician Fading Model

Rician Fading is the addition to all the normal multipaths a direct LOS path.
JN
ll
A

Figure 5.8: Ricean probability density function.

93
r −(r2 +A2 ) Ar
2σ2
I( ) (5.58)
fR(r) = e
σ2 σ2 0

for all A ≥ 0 and r ≥ 0. Here A is the peak amplitude of the dominant signal and

I0(.) is the modified Bessel function of the first kind and zeroth order.
A factor K is defined as
A2

ld
K dB = 10 log (5.59)
2σ2
As A → 0 then KdB → ∞.

Generalized Model: Nakagami Distribution

or
A generalization of the Rayleigh and Rician fading is the Nakagami distribution.

W
TU

Figure 5.9: Nakagami probability density function.

Its pdf is given as,


JN

2rm−1 mm −mr2
fR(r) = ( )e Ω (5.60)
Γ(m) Ωm
where,
Γ(m) is the gamma function
Ω is the average signal power and
m is the fading factor.It is always greater than or equal to 0.5.
ll

When m=1, Nakagami model is the Rayleigh model.


A

When
(M + 1)2
m=
2M + 1

94
ld
or
Figure 5.10: Schematic representation of level crossing with a Rayleigh fading enve-
lope at 10 Hz Doppler spread.

where

Nakagami fading is the Rician fading.


W M=
A

As m → ∞ Nakagami fading is the impulse channel and no fading occurs.


TU

Second Order Statistics

To design better error control codes, we have two important second order param-
eters of fading model, namely the level crossing rate (LCR) and average fade
duration (AFD). These parameters can be utilized to assess the speed of the user
JN

by measuring them through the reverse channel. The LCR is the expected rate at
which the Rayleigh fading envelope normalized to the local rms amplitude crosses a
specific level ’R’ in a positive going direction.
¸∞ √ 2
NR = ṙp(R, ṙ)dṙ = 2πfD ρe−ρ (5.61)
0
ll

where r˙ is the time derivative of r(t), fD is the maximum Doppler shift and ρ is the
value of the specified level R, normalized to the local rms amplitude of the fading
A

envelope.
The other important parameter, AFD, is the average period time for which the

95
receiver power is below a specified level R.
1
τ¯ = Pr(r ≤ R) (5.62)
Nr

As
¸R
Pr (r ≤ R) = p(r)dr = 1 − e−ρ ,
2
(5.63)
0

ld
therefore, 2
1 − e−ρ
(5.64)
τ¯ = √ D ρe−ρ
2
2πf
2 − 1

e− .ρ (5.65)
= √2πf ρ

or
D

Apart from LCR, another parameter is fading rate, which is defined as the number of
times the signal envelope crosses the middle value (rm) in a positive going direction
per unit time. The average rate is expressed as

W
N (rm ) =
2v
λ
.

Another statistical parameter, sometimes used in the mobile communication, is


called as depth of fading.
(5.66)

It is defined as the ratio between the minimum value


TU
and the mean square value of the faded signal. Usually, an average value of 10% as
depth of fading gives a marginal fading scenario.

Simulation of Rayleigh Fading Models


JN

Clarke’s Model: without Doppler Effect

In it, two independent Gaussian low pass noise sources are used to produce in-phase
and quadrature fading branches. This is the basic model and is useful for slow fading
channel. Also the Doppler effect is not accounted for.
ll

Clarke and Gans’ Model: with Doppler Effect

In this model, the output of the Clarke’s model is passed through Doppler filter in
A

the RF or through two initial baseband Doppler filters for baseband processing as
shown in Figure 5.11. Here, the obtained Rayleigh output is flat faded signal but
not frequency selective.

96
ld
or
W
Figure 5.11: Clarke and Gan’s model for Rayleigh fading generation using quadra-
ture amplitude modulation with (a) RF Doppler filter, and, (b) baseband Doppler
TU
filter.

Rayleigh Simulator with Wide Range of Channel Conditions

To get a frequency selective output we have the following simulator through which
both the frequency selective and flat faded Rayleigh signal may be obtained. This
JN

is achieved through varying the parameters ai and τi, as given in Figure 5.12.

Two-Ray Rayleigh Faded Model

The above model is, however, very complex and difficult to implement. So, we have
the two ray Rayleigh fading model which can be easily implemented in software as
ll

shown in Figure 5.13.

hb (t) = α1 ejφ1 δ(t) + α2 ejφ2 δ(t − τ ) (5.67)


A

where α1 and α2 are independent Rayleigh distributed and φ1 and φ2 are indepen-
dent and uniformaly distributed over 0 to 2π. By varying τ it is possible to create
a wide range of frequency selective fading effects.

97
ld
or
W
TU

Figure 5.12: Rayleigh fading model to get both the flat and frequency selective
channel conditions.

Saleh and Valenzuela Indoor Statistical Model


JN

This method involved averaging the square law detected pulse response while sweep-
ing the frequency of the transmitted pulse. The model assumes that the multipath
components arrive in clusters. The amplitudes of the received components are in-
dependent Rayleigh random variables with variances that decay exponentially with
cluster delay as well as excess delay within a cluster. The clusters and multipath
ll

components within a cluster form Poisson arrival processes with different rates.
A

SIRCIM/SMRCIM Indoor/Outdoor Statistical Models

SIRCIM (Simulation of Indoor Radio Channel Impulse-response Model) generates


realistic samples of small-scale indoor channel impulse response measurements. Sub-

98
ld
Figure 5.13: Two-ray Rayleigh fading model.

or
sequent work by Huang produced SMRCIM (Simulation of Mobile Radio Channel
Impulse-response Model), a similar program that generates small-scale urban cellu-
lar and micro-cellular channel impulse responses.

Conclusion
W
In this chapter, the main channel impairment, i.e., fading, has been introduced which
becomes so severe sometimes that even the large scale path loss becomes insignificant
TU
in comparison to it. Some statistical propagation models have been presented based
on the fading characteristics. Mainly the frequency selective fading, fast fading and
deep fading can be considered the major obstruction from the channel severity view
point. Certain efficient signal processing techniques to mitigate these effects will be
JN

discussed in Chapter 7.

References

1. T. S. Rappaport, Wireless Communications: Principles and Practice, 2nd ed.


Singapore: Pearson Education, Inc., 2002.
ll

2. S. Haykin and M. Moher, Modern Wireless Communications. Singapore: Pear-


son Education, Inc., 2002.
A

3. J. W. Mark and W. Zhuang, Wireless Communications and Networking. New


Delhi: PHI, 2005.

99
4. K. Feher, Wireless Digital Communications: Modulation and Spread Spectrum
Applications. Upper Saddle River, NJ: Prentice Hall, 1995.

5. R. Blake, Wireless Communications Technology. Delmar, Singapore: Thom-


son Asia Pvt Ltd, 2004.

6. D. P. Agarwal and Q-A. Zeng, Introduction to Wireless and Mobile Systems.

ld
Nelson, India: Thomson Learning, 2007.

or
W
TU
JN
ll
A

100
Chapter 6

ld
Transmitter and Receiver

or
Techniques

Introduction
W
Electrical communication transmitter and receiver techniques strive toward obtain-
ing reliable communication at a low cost, with maximum utilization of the channel
resources. The information transmitted by the source is received by the destina-
TU

tion via a physical medium called a channel. This physical medium, which may be
wired or wireless, introduces distortion, noise and interference in the transmitted
information bearing signal. To counteract these effects is one of the requirements
while designing a transmitter and receiver end technique. The other requirements
JN

are power and bandwidth efficiency at a low implementation complexity.

Modulation

Modulation is a process of encoding information from a message source in a man-


ner suitable for transmission. It involves translating a baseband message signal to
ll

a passband signal. The baseband signal is called the modulating signal and the
passband signal is called the modulated signal. Modulation can be done by varying
A

certain characteristics of carrier waves according to the message signal. Demodu-


lation is the reciprocal process of modulation which involves extraction of original
baseband signal from the modulated passband signal.

101
Choice of Modulation Scheme

Several factors influence the choice of a digital modulation scheme. A desirable


modulation scheme provides low bit error rates at low received signal to noise ratios,
performs well in multipath and fading conditions, occupies a minimum of bandwidth,
and is easy and cost-effective to implement. The performance of a modulation

ld
scheme is often measured in terms of its power efficiency and bandwidth efficiency.
Power efficiency describes the ability of a modulation technique to preserve the
fidelity of the digital message at low power levels. In a digital communication system,
in order to increase noise immunity, it is necessary to increase the signal power.

or
Bandwidth efficiency describes the ability of a modulation scheme to accommodate
data within a limited bandwidth.
The system capacity of a digital mobile communication system is directly related

W
to the bandwidth efficiency of the modulation scheme, since a modulation with a
greater value of ηb(= RB) will transmit more data in a given spectrum allocation.
There is a fundamental upper bound on achievable bandwidth efficiency. Shan-
non’s channel coding theorem states that for an arbitrarily small probability of error,
the maximum possible bandwidth efficiency is limited by the noise in the channel,
TU
and is given by the channel capacity formula
C S
ηBmax = = log2(1 + ) (6.1)
B N

Advantages of Modulation
JN

1. Facilitates multiple access: By translating the baseband spectrum of signals


from various users to different frequency bands, multiple users can be accom-
modated within a band of the electromagnetic spectrum.

2. Increases the range of communication: Low frequency baseband signals suffer


from attenuation and hence cannot be transmitted over long distances. So
ll

translation to a higher frequency band results in long distance transmission.


A

3. Reduction in antenna size: The antenna height and aperture is inversely pro-
portional to the radiated signal frequency and hence high frequency signal
radiation result in smaller antenna size.

102
Linear and Non-linear Modulation Techniques

The mathematical relation between the message signal (applied at the modulator
input) and the modulated signal (obtained at the modulator output) decides whether
a modulation technique can be classified as linear or non-linear. If this input-output
relation satisfies the principle of homogeneity and superposition then the modulation

ld
technique is said to be linear. The principle of homogeneity states that if the input
signal to a system (in our case the system is a modulator) is scaled by a factor then
the output must be scaled by the same factor. The principle of superposition states
that the output of a linear system due to many simultaneously applied input signals

or
is equal to the summation of outputs obtained when each input is applied one at a
time.
For example an amplitude modulated wave consists of the addition two terms: the

Then,
W
message signal multiplied with the carrier and the carrier itself. If m(t) is the
message signal and sAM (t) is the modulated signal given by:

sAM (t) = Ac[1 + km(t)] cos(2πfct) (6.2)


TU

1. From the principle of homogeneity: Let us scale the input by a factor a. So

m(t) = am1(t) and the corresponding output becomes :

sAM 1(t) = Ac [1 + am1(t)] cos(2πfc t) (6.3)


ƒ= asAM 1 (t)
JN

2. From the principle of superposition: Let m(t) = m1 (t) + m2(t) be applied


simultaneously at the input of the modulator. The resulting output is:

sAM (t) = Ac[1 + m1(t) + m2(t)] cos(2πfct) (6.4)


ƒ= sAM 1 (t) + sAM 2 (t)
ll

= Ac[2 + m1(t) + m2(t)] cos(2πfct)

Here, sAM 1(t) and sAM 2(t) are the outputs obtained when m1(t) and m2(t)
A

are applied one at a time.

Hence AM is a nonlinear technique but DSBSC modulation is a linear technique


since it satisfies both the above mentioned principles.

103
Amplitude and Angle Modulation

Depending on the parameter of the carrier (amplitude or angle) that is changed


in accordance with the message signal, a modulation scheme can be classified as
an amplitude or angle modulation. Amplitude modulation involves variation of
amplitude of the carrier wave with changes in the message signal. Angle modulation

ld
varies a sinusoidal carrier signal in such a way that the angle of the carrier is varied
according to the amplitude of the modulating baseband signal.

Analog and Digital Modulation Techniques

or
The nature of the information generating source classifies a modulation technique as
an analog or digital modulation technique. When analog messages generated from
a source passe through a modulator, the resulting amplitude or angle modulation

W
technique is called analog modulation. When digital messages undergo modulation
the resulting modulation technique is called digital modulation.

Signal Space Representation of Digitally Modulated


TU
Signals

Any arbitrary signal can be expressed as the linear combination of a set of orthog-
onal signals or equivalently as a point in an M dimensional signal space, where M
denotes the cardinality of the set of orthogonal signals. These orthogonal signals are
JN

normalized with respect to their energy content to yield an orthonormal signal set
having unit energy. These orthonormal signals are independent of each other and
form a basis set of the signal space.
Generally a digitally modulated signal s(t), having a symbol duration T, is ex-

pressed as a linear combination of two orthonormal signals φ1(t) and φ2(t), consti-
tuting the two orthogonal axis in this two dimensional signal space and is expressed
ll

mathematically as,
A

s(t) = s1 φ1(t) + s2φ 2(t) (6.5)

where φ1(t) and φ2(t) are given by,


.
φ1(t) = 2 cos(2πfct) (6.6)
T

104
.
φ2(t) = 2 cos(2πfct) (6.7)
T
The coefficients s1 and s2 form the coordinates of the signal s(t) in the two dimen-
sional signal space.

Complex Representation of Linear Modulated Sig-

ld
nals and Band Pass Systems

A band-pass signal s(t) can be resolved in terms of two sinusoids in phase quadrature

or
as follows:

s(t) = sI (t)cos(2πfct) − sQ(t)sin(2πfc t) (6.8)

Hence sI (t) and sQ(t) are known as the in-phase and quadrature-phase components

W
respectively. When sI (t) and sQ(t) are incorporated in the formation of the following
complex signal,

s̃(t) = sI (t) + sQ (t)

then s(t) can be expressed in a more compact form as:


(6.9)
TU
s(t) = Re{s̃(t)e(j 2πfc t) } (6.10)

where s̃(t) is called the complex envelope of s(t).


Analogously, band-pass systems characterized by an impulse response h(t) can
be expressed in terms of its in-phase and quadrature-phase components as:
JN

h(t) = hI (t)cos(2πfc t) − hQ (t)sin(2πfc t) (6.11)

The complex baseband model for the impulse response therefore becomes,

h̃(t) = hI (t) + hQ (t) (6.12)

h(t) can therefore be expressed in terms of its complex envelope as


ll

h(t) = Re{h̃(t)ej 2πfc t }. (6.13)


A

When s(t) passes through h(t), then in the complex baseband domain, the output
r̃(t) of the bandpass system is given by the following convolution

r̃(t) = 1 s̃(t) ⊗ h̃(t) (6.14)


2

105
Linear Modulation Techniques

Amplitude Modulation (DSBSC)

Generally, in amplitude modulation, the amplitude of a high frequency carrier signal,


cos(2πfct), is varied in accordance to the instantaneous amplitude of the modulat-
ing message signal m(t). The resulting modulated carrier or AM signal can be

ld
represented as:
sAM (t) = Ac[1 + km(t)] cos(2πfct). (6.15)

The modulation index k of an AM signal is defined as the ratio of the peak message

or
signal amplitude to the peak carrier amplitude. For a sinusoidal modulating signal
Am
m(t) = Ac cos(2πfmt), the modulation index is given by
A
k = m. (6.16)
Ac

W
This is a nonlinear technique and can be made linear by multiplying the carrier with
the message signal.The resulting modulation scheme is known as DSBSC modula-
tion. In DSBSC the amplitude of the transmitted signal, s(t), varies linearly with
the modulating digital signal, m(t). Linear modulation techniques are bandwidth
TU
efficient and hence are very attractive for use in wireless communication systems
where there is an increasing demand to accommodate more and more users within
a limited spectrum. The transmitted signal DSBSC signal s(t) can be expressed as:

s(t) = Am(t)exp(j2πfct). (6.17)


JN

If m(t) is scaled by a factor of a, then s(t), the output of the modulator, is also
scaled by the same factor as seen from the above equation. Hence the principle of
homogeneity is satisfied. Moreover,

s12 (t) = A[m1 (t) + m2 (t)]cos(2πfc t) (6.18)

= Am1 (t)cos(2πfc t) + Am2 (t)cos(2πfc t)


ll

= s1(t) + s2(t)

where A is the carrier amplitude and fc is the carrier frequency. Hence the principle
A

of superposition is also satisfied. Thus DSBSC is a linear modulation technique.


AM demodulation techniques may be broadly divided into two categories: co-
herent and non-coherent demodulation. Coherent demodulation requires knowledge

106
ld
Figure 6.1: BPSK signal constellation.

or
of the transmitted carrier frequency and phase at the receiver, whereas non-coherent
detection requires no phase information.

BPSK

W
In binary phase shift keying (BPSK), the phase of a constant amplitude carrier
signal is switched between two values according to the two possible signals m1 and
m2 corresponding to binary 1 and 0, respectively. Normally, the two phases are
TU
separated by 180o. If the sinusoidal carrier has an amplitude A, and energy per bit
1 2
Eo = A Tb then the transmitted BPSK signal is
2 c
.
2Eb
sBP SK (t) = m(t) cos(2πfct + θc ). (6.19)
Tb
JN

A typical BPSK signal constellation diagram is shown in Figure 6.1.


The probability of bit error for many modulation schemes in an AWttN channel
is found using the Q-function of the distance between the signal points. In case of
BPSK, .
2Eb
PeBP SK = Q( ). (6.20)
N0
ll

QPSK

The Quadrature Phase Shift Keying (QPSK) is a 4-ary PSK signal. The phase of
A

the carrier in the QPSK takes 1 of 4 equally spaced shifts. Although QPSK can
be viewed as a quaternary modulation, it is easier to see it as two independently
modulated quadrature carriers. With this interpretation, the even (or odd) bits are

107
ld
or
Figure 6.2: QPSK signal constellation.

W
TU

Figure 6.3: QPSK transmitter.


JN

used to modulate the in-phase component of the carrier, while the odd (or even)
bits are used to modulate the quadrature-phase component of the carrier.
The QPSK transmitted signal is defined by:

si(t) = A cos(ωt + (i − 1)π/2), i = (1, 2, 3, 4) (6.21)


ll

and the constellation disgram is shown in Figure 6.2.


A

Offset-QPSK

As in QPSK, as shown in Figure 6.3, the NRZ data is split into two streams of odd
and even bits. Each bit in these streams has a duration of twice the bit duration,

108
ld
or
Figure 6.4: DQPSK constellation diagram.

T b, of the original data stream. These odd (d1(t)) and even bit streams (d2(t)) are
then used to modulate two sinusoidals in phase quadrature,and hence these data

W
streams are also called the in-phase and and quadrature phase components. After
modulation they are added up and transmitted. The constellation diagram of Offset-
QPSK is the same as QPSK. Offset-QPSK differs from QPSK in that the d1(t) and
d2(t) are aligned such that the timing of the pulse streams are offset with respect
TU
to each other by Tb seconds. From the constellation diagram it is observed that a
signal point in any quadrant can take a value in the diagonally opposite quadrant
only when two pulses change their polarities together leading to an abrupt 180 degree
phase shift between adjacent symbol slots. This is prevented in O-QPSK and the
allowed phase transitions are ± 90 degree.
JN

Abrupt phase changes leading to sudden changes in the signal amplitude in the
time domain corresponds to significant out of band high frequency components in
the frequency domain. Thus to reduce these sidelobes spectral shaping is done at
baseband. When high efficiency power amplifiers, whose non-linearity increases as
the efficiency goes high, are used then due to distortion, harmonics are generated
ll

and this leads to what is known as spectral regrowth. Since sudden 180 degree phase
changes cannot occur in OQPSK, this problem is reduced to a certain extent.
A

109
π/4 DQPSK

The data for π/4 DQPSK like QPSK can be thought to be carried in the phase of a
single modulated carrier or on the amplitudes of a pair of quadrature carriers. The
modulated signal during the time slot of kT < t < (k + 1)T given by:

s(t) = cos(2πfct + ψk+1) (6.22)

ld
Here, ψk+1 = ψk + ∆ψk and ∆ψk can take values π/4 for 00, 3π/4 for 01, −3π/4
for 11 and −π/4 for 10. This corresponds to eight points in the signal constellation
but at any instant of time only one of the four points are possible: the four points

or
on axis or the four points off axis. The constellation diagram along with possible
transitions are shown in Figure 6.4.

Line Coding

W
Specific waveforms are required to represent a zero and a one uniquely so that a
sequence of bits is coded into electrical pulses. This is known as line coding. There
are various ways to accomplish this and the different forms are summarized below.
TU

1. Non-return to zero level (NRZ-L): 1 forces a a high while 0 forces a low.

2. Non-return to zero mark (NRZ-M): 1 forces negative and positive transitions


while 0 causes no transitions.
JN

3. Non-return to zero space (NRZ-S): 0 forces negative and positive transitions


while 1 causes no transitions.

4. Return to zero (RZ): 1 goes high for half a period while 0 remains at zero
state.

5. Biphase-L: Manchester 1 forces positive transition while 0 forces negative tran-


ll

sition. In case of consecutive bits of same type a transition occurs in the


beginning of the bit period.
A

6. Biphase-M: There is always a transition in the beginning of a bit interval. 1


forces a transition in the middle of the bit while 0 does nothing.

110
ld
or
W
TU

Figure 6.5: Scematic of the line coding techniques.

7. Biphase-S: There is always a transition in the beginning of a bit interval. 0


forces a transition in the middle of the bit while 1 does nothing.
JN

8. Differential Manchester: There is always a transition in the middle of a bit


interval. 0 forces a transition in the beginning of the bit while 1 does nothing.

9. Bipolar/Alternate mark inversion (AMI): 1 forces a positive or negative pulse


for half a bit period and they alternate while 0 does nothing.
ll

All these schemes are shown in Figure 6.5.


A

Pulse Shaping

Let us think about a rectangular pulse as defined in BPSK. Such a pulse is not
desirable for two fundamental reasons:

111
ld
Figure 6.6: Rectangular Pulse

or
(a) the spectrum of a rectangular pulse is infinite in extent. Correspondingly, its
frequency content is also infinite. But a wireless channel is bandlimited, means it
would introduce signal distortion to such type of pulses,

W
(b) a wireless channel has memory due to multipath and therefore it introduces ISI.
In order to mitigate the above two effects, an efficient pulse shaping funtion or
a premodulation filter is used at the Tx side so that QoS can be maintained to the
mobile users during communication. This type of technique is called pulse shaping
TU
technique. Below, we start with the fundamental works of Nyquist on pulse shaping
and subsequently, we would look into another type of pulse shaping technique.

Nyquist pulse shaping

There are a number of well known pulse shaping techniques which are used to simul-
JN

taneously to reduce the inter-symbol effects and the spectral width of a modulated
digital signal. We discuss here about the fundamental works of Nyquist. As pulse
shaping is difficult to directly manipulate the transmitter spectrum at RF frequen-
cies, spectral shaping is usually done through baseband or IF processing.
Let the overall frequency response of a communication system (the transmitter,
ll

channel and receiver) be denoted as Hef f (f ) and according to Nyquist it must be


given by:
1 f
A

Hef f (f ) = rect( ) (6.23)


fs fs
Hence, the ideal pulse shape for zero ISI, given by hef f (t), such that,

Hef f (f ) ↔ hef f (t) (6.24)

112
ld
or
Figure 6.7: Raised Cosine Pulse.

is given by:
sin( πt )

W
hef f (t) = πt Ts

Raised Cosine Roll-Off Filtering


Ts
(6.25)

(6.26)

1
If we take a rectangular filter with bandwidth f0 ≥
TU
2Ts and convolve it with any
arbitrary even function Z(f) with zero magnitude outside the passband of the rect-
angular filter then a zero ISI effect would be achieved. Mathematically,
f
Hef f (f ) = rect( ) ∗ Z(f ), (6.27)
f0
sin( πt )
JN

Ts
hef f (t) = πt
z(t), (6.28)
Ts
cos(πρt/Ts)
z(t) = . (6.29)
1 − (∆ρt/2T s)2

with ρ being the roll off factor ∈ [0, 1]. As ρ increases roll off in frequency domain
increases but that in time domain decreases.
ll

Realization of Pulse Shaping Filters


A

Since hef f (t) is non-causal, pulse shaping filters are usually truncated within ±6Ts
about t = 0 for each symbol. Digital communication systems thus often store several
symbols at a time inside the modulator and then clock out a group of symbols by

113
using a look up table that represents discrete time waveforms of stored symbols.
This is the way to realize the pulse shaping filters using real time processors.
Non-Nyquist pulse shaping are also useful, which would be discussed later in this
chapter while discussing GMSK.

Nonlinear Modulation Techniques

ld
Many practical mobile radio communications use nonlinear modulation methods,where
the amplitude of the carrier is constant,regardless of the variations in the modulating

or
signal.The Constant envelope family of modulations has the following advantages :

1. Power efficient class C amplifiers without introducing degradation in the spec-


tral occupancy of the transmitted signal.

W
2. Low out-of-band radiation of the order of -60 dB to -70dB can be achieved.

3. Limiter-discriminator detection can be used,which simplifies receiver design


and provides high immunity against random FM noise and signal fluctuations
due to Rayleigh fading.
TU

However, even if constant envelope has many advantages it still uses more BW
than linear modulation schemes.

Angle Modulation (FM and PM)


JN

There are a number of ways in which the phase of a carrier signal may be varied
in accordance with the baseband signal; the two most important classes of angle
modulation being frequency modulation and phase modulation.
Frequency modulation (FM) involves changing of the frequency of the carrier
signal according to message signal. As the information in frequency modulation is
in the frequency of modulated signal, it is a nonlinear modulation technique. In this
ll

method, the amplitude of the carrier wave is kept constant (this is why FM is called
constant envelope). FM is thus part of a more general class of modulation known
A

as angle modulation.
Frequency modulated signals have better noise immunity and give better perfor-
mance in fading scenario as compared to amplitude modulation.Unlike AM, in an

114
FM system, the modulation index, and hence bandwidth occupancy, can be varied
to obtain greater signal to noise performance.This ability of an FM system to trade
bandwidth for SNR is perhaps the most important reason for its superiority over
AM. However, AM signals are able to occupy less bandwidth as compared to FM
signals, since the transmission system is linear.
An FM signal is a constant envelope signal, due to the fact that the envelope of

ld
the carrier does not change with changes in the modulating signal. The constant
envelope of the transmitted signal allows efficient Class C power amplifiers to be
used for RF power amplification of FM. In AM, however, it is critical to maintain

or
linearity between the applied message and the amplitude of the transmitted signal,
thus linear Class A or AB amplifiers, which are not as power efficient, must be used.
FM systems require a wider frequency band in the transmitting media (generally
several times as large as that needed for AM) in order to obtain the advantages of

W
reduced noise and capture effect. FM transmitter and receiver equipment is also
more complex than that used by amplitude modulation systems. Although frequency
modulation systems are tolerant to certain types of signal and circuit nonlinearities,
special attention must be given to phase characteristics. Both AM and FM may be
TU
demodulated using inexpensive noncoherent detectors. AM is easily demodulated
using an envelope detector whereas FM is demodulated using a discriminator or
slope detector. In FM the instantaneous frequency of the carrier signal is varied
linearly with the baseband message signal m(t), as shown in following equation:
¸
JN

sF M (t) = Ac cos[2πfc t + θ(t)] = Ac cos[2πfc t + 2πkf m(η)dη] (6.30)

where Ac, is the amplitude of the carrier, fc is the carrier frequency, and kf is the
frequency deviation constant (measured in units of Hz/V).
Phase modulation (PM) is a form of angle modulation in which the angle θ(t) of
the carrier signal is varied linearly with the baseband message signal m(t), as shown
ll

in equation below.
sP M (t) = Ac cos(2πfc t + kθ m(t)) (6.31)
A

The frequency modulation index βf , defines the relationship between the message
amplitude and the bandwidth of the transmitted signal, and is given by
kf Am ∆
βf = = (6.32)
W W

115
where Am is the peak value of the modulating signal, ∆f is the peak frequency
deviation of the transmitter and W is the maximum bandwidth of the modulating
signal.
The phase modulation index βp is given by

βp = kθ Am = ∆θ (6.33)

ld
where, ∆θ is the peak phase deviation of the transmitter.

BFSK

or
In Binary Frequency Shift keying (BFSK),the frequency of constant amplitude car-
rier signal is switched between two values according to the two possible message
states (called high and low tones) corresponding to a binary 1 or 0. Depending on

S(t) =
.
W
how the frequency variations are imparted into the transmitted waveform,the FSK
signal will have either a discontinuous phase or continuous phase between bits. In
general, an FSK signal may be represented as

(2Eb/T ) cos(2πf it). (6.34)


TU
where T is the symbol duration and Eb is the energy per bit.
.
Si = (Eb)φ(t). (6.35)
.
φ(t) = /T ) cos(2πf t).
(2 i (6.36)
JN

There are two FSK signals to represent 1 and 0, i.e.,


.
S1 (t) = (2Eb /T ) cos(2πf1 t + θ(0)) →1 (6.37)
.
S2 (t) = (2Eb /T ) cos(2πf2 t + θ(0)) →0 (6.38)

where θ(0) sums the phase up to t = 0. Let us now consider a continuous phase
ll

FSK as
.
S(t) = E /T
(2 b
) cos(2πf ct + θ(t)). (6.39)
A

Expressing θ(t) in terms of θ(0) with a new unknown factor h, we get

θ(t) = θ(0) ± πht/T 0 ≤ t≤ T (6.40)

116
and therefore
. .
2Eb 2Eb cos(2π(f ± h/2T )t + θ(0)).(6.41)
S(t) = cos(2πfct ± πht/T + θ(0)) = c
T T

It shows that we can choose two frequencies f1 and f2 such that

f1 = fc + h/2T (6.42)

ld
f2 = fc − h/2T (6.43)

for which the expression of FSK conforms to that of CPFSK. On the other hand, fc

or
and h can be expressed in terms of f1 and f2 as

fc = [f1 + f2 ]/2 (6.44)


(f1 − f2 )
h= . (6.45)
1/T

W
Therefore, the unknown factor h can be treated as the difference between f1 and f2,
normalized with respect to bit rate 1/T . It is called the deviation ratio. We know
that θ(t) − θ(0) = ±πht/T , 0 ≤ t ≤ T . If we substitute t = T , we have

θ(T ) − θ(0) = ±πh where (6.46)


TU
= πh →1 (6.47)

= −πh →0 (6.48)

This type of CPFSK is advantageous since by looking only at the phase, the trans-
mitted bit can be predicted. In Figure 6.8, we show a phase tree of such a CPFSK
JN

signal with the transmitted bit stream of 1101000.


A special case of CPFSK is achieved with h = 0.5, and the resulting scheme is
called Minimum Shift Keying (MSK) which is used in mobile communications. In
this case, the phase differences reduce to only ±π/2 and the phase tree is called the
phase trellis. An MSK signal can also be thought as a special case of OQPSK where
ll

the baseband rectangular pulses are replaced by half sinusoidal pulses. Spectral
characteristics of an MSK signal is shown in Figure 6.9 from which it is clear that
A

ACI is present in the spectrum. Hence a pulse shaping technique is required. In


order to have a compact signal spectrum as well as maintaining the constant envelope
property, we use a pulse shaping filter with

117
ld
or
Figure 6.8: Phase tree of 1101000 CPFSK sequence.

W
TU

Figure 6.9: Spectrum of MSK


JN

1. a narrow BW frequency and sharp cutoff characteristics (in order to suppress


the high frequency component of the signal);

2. an impulse response with relatively low overshoot (to limit FM instant fre-
quency deviation;

3. a phase trellis with ±π/2 for odd T and 0 or π values for even T.
ll

GMSK Scheme
A

GMSK is a simple modulation scheme that may be taken as a derivative of MSK.


In GMSK, the sidelobe levels of the spectrum are further reduced by passing a non-

118
ld
or
Figure 6.10: GMSK generation scheme.

W
return to zero (NRZ-L) data waveform through a premodulation Gaussian pulse
shaping filter. Baseband Gaussian pulse shaping smoothes the trajectory of the
MSK signals and hence stabilizes instantaneous frequency variations over time. This
has the effect of considerably reducing the sidelobes in the transmitted spectrum.
A GMSK generation scheme with NRZ-L data is shown in Figure 6.10 and a receiver
TU

of the same scheme with some MSI gates is shown in Figure 6.11.

GMSK Generator

The GMSK premodulation filter has characteristic equation given by


JN

H(f ) = exp(−(ln 2/2)(f/B)2 ) (6.49)

H(f ) = exp(−(αf ) 2)

where,
(α)2 = ln 2/2(1/B)2 . (6.50)
ll

The premodulation Gaussian filtering introduces ISI in the transmitted signal, but
it can be shown that the degradation is not that great if the 3dB bandwidth-bit
A

duration product (BT) is greater than 0.5.


Spectrum of GMSK scheme is shown in Figure 6.12. From this figure, it is evident
that when we are decreasing BT product, the out of band response decreases but

119
ld
or
W
Figure 6.11: A simple GMSK receiver.

on the other hand irreducible error rate of the LPF for ISI increases. Therefore, a
TU

compromise between these two is required.


Problem: Find the 3dB BW for a Gaussian LPF used to produce 0.25 GMSK
with a channel data rate Rb=270 kbps.What is the 90 percent power BW of the RF
filter?
JN

Solution: From the problem statement it is clear that

T = 1/Rb = 1/270 ∗ (103) = 3.7µsec (6.51)

Solving for B where BT = 0.25,

B = 0.25/T = 67.567kHz (6.52)


ll

Thus the 3 - dB bandwidth is 67.567 kHz. We use below table fig 6 to find out that
90 % power bandwidth is 0.57 Rb.
A

90 % RF BW = 0.57Rb = 153.9 kHz.

120
ld
or
Figure 6.12: Spectrum of GMSK scheme.

W
Two Practical Issues of Concern

Inter Channel Interference

In FDMA, subscribers are allotted frequency slots called channels in a given band
TU

of the electromagnetic spectrum. The side lobes generated due to the transmission
of a symbol in a particular channel overlaps with the channels placed adjacently.
This is because of the fact that transmission of a time limited pulse leads to spectral
spreading in the frequency domain. During simultaneous use of adjacent channels,
JN

when there is significant amount of power present in the side lobes, this kind of
interference becomes so severe that the required symbol in a particular frequency
slot is completely lost.
Moreover if two terminals transmit equal power then due to wave propagation
through different distances to the receiver, the received signal levels in the two
frequency slots will differ greatly. In such a case the side lobes of the stronger signal
ll

will severely degrade the transmitted signal in the next frequency slot having low
power level. This is known as the near far problem.
A

121
Power Amplifier Nonlinearity

Power amplifiers may be designed as class A, class B, class AB, class C and class D.
They form an essential section of mobile radio terminals. Due to power constraints
on a transmitting terminal, an efficient power amplifier is required which can convert
most of the input power to RF power. Class A amplifier is a linear amplifier but

ld
it has a power efficiency of only 25 %. As we go for subsequent amplifiers having
greater power efficiency, the nonlinearity of the amplifier increases.
In general, an amplifier has linear input output characteristics over a range
of input signal level, that is, it has a constant gain. However, beyond an input

or
threshold level, the gain of the amplifier starts decreasing. Thus the amplitude of
a signal applied at the input of an amplifier suffers from amplitude distortion and
the resulting waveform obtained at the output of the amplifier is of the form of

form of phase modulation. W


an amplitude modulated signal. Similarly, the phase characteristic of a practical
amplifier is not constant over all input levels and results in phase distortion of the

The operating point of a practical amplifier is given in terms of either the input
back-off or the output back-off.
TU
. .
Vin,rms
Input back − off = 10 log1 0 (6.53)
Vout,rms
. .
Vout,rms
Output back − off = 10 log1 0 (6.54)
Vout,rms
JN

Receiver performance in multipath channels

For a flat fading channel, the probability of error for coherent BPSK and coherent
BFSK are respectively given as,
. .
1 γ . (6.55)
Pe,BP SK = 1− 1+γ
ll

2 . .
1 γ . (6.56)
Pe,BF SK = 1 − 2 +γ
2
A

(6.57)

where γ is given by,


Eb
γ= E(α 2) (6.58)
N0

122
α2 represents the instantaneous power values of the Rayleigh fading channel and E
denotes the expectation operator.
Similarly, for differential BPSK and non coherent BFSK probability of error
expressions are
1
Pe,DP SK = (6.59)
2(11+ γ)

ld
Pe,N CF SK = . (6.60)
(2 + γ)
For large values of SNR = Eb the error probability given above have the simplified
N0

expression.

or
1
Pe,BP SK = (6.61)

1
Pe,BF SK = (6.62)

1

W
Pe,DP SK =

Pe,N CFSK = .
1

γ
(6.63)

(6.64)

From the above equations we observe that an inverse algebraic relation exists be-
tween the BER and SNR. This implies that if the required BER range is around
TU

10−3 to 10−6, then the SNR range must be around 30dB to 60dB.

Bit Error Rate and Symbol Error Rate

Bit error rate (Peb) is the same as symbol error rate (Pes) when a symbol consists
JN

of a single bit as in BPSK modulation. For an MPSK scheme employing gray coded
modulation, where N bits are mapped to a one of the M symbols, such that 2 N = M ,
Peb is given by Pes
P (6.65)
eb ≈
log2M
And for M-ary orthogonal signalling Peb is given by
M/2
ll

Pes. (6.66)
Peb = 1
M−
A

Example of a Multicarrier Modulation: OFDM

Multiplexing is an important signal processing operation in which a number of sig-


nals are combined and transmitted parallelly over a common channel. In order to

123
avoid interference during parallel transmission, the signals can be separated in fre-
quency and then the resulting technique is called Frequency Division Multiplexing
(FDM). In FDM, the adjacent bands are non overlapping but if overlap is allowed by
transmitting signals that are mutually orthogonal (that is, there is a precise math-
ematical relationship between the frequencies of the transmitted signals) such that
one signal has zero effect on another, then the resulting transmission technique is

ld
known as Orthogonal Frequency Division Multiplexing (OFDM).
OFDM is a technique of transmitting high bit rate data into several parallel
streams of low bit rate data. At any instant, the data transmitted simultaneously

or
in each of these parallel data streams is frequency modulated by carriers (called
subcarriers) which are orthogonal to each other. For high data rate communication
the bandwidth (which is limited) requirement goes on increasing as the data rate
increases or the symbol duration decreases. Thus in OFDM, instead of sending a

W
particular number of symbols, say P, in T seconds serially, the P symbols can be
sent in parallel with symbol duration now increased to T seconds instead of T/P
seconds as was previously.
This offers many advantages in digital data transmission through a wireless time
TU
varying channel. The primary advantage of increasing the symbol duration is that
the channel experiences flat fading instead of frequency selective fading since it is
ensured that in the time domain the symbol duration is greater than the r.m.s.
delay spread of the channel. Viewed in the frequency domain this implies that the
bandwidth of the OFDM signal is less than coherent bandwidth of the channel.
JN

Although the use of OFDM was initially limited to military applications due to
cost and complexity considerations, with the recent advances in large-scale high-
speed DSP, this is no longer a major problem. This technique is being used, in
digital audio broadcasting (DAB), high definition digital television broadcasting
(HDTV), digital video broadcasting terrestrial TV (DVB-T), WLAN systems based
ll

on IEEE 802.11(a) or HiperLan2, asymmetric digital subscriber lines (ADSL) and


mobile communications. Very recently, the significance of the COFDM technique for
A

UWA (underwater acoustic channel) has also been indicated. Moreover related or
combined technology such as CDMA-OFDM, TDMA-OFDM, MIMO-OFDM, Vec-
tor OFDM (V-OFDM), wide-band OFDM (W-OFDM), flash OFDM (F-OFDM),

124
OFDMA, wavelet-OFDM have presented their great advantages in certain applica-
tion areas.

Orthogonality of Signals

Orthogonal signals can be viewed in the same perspective as we view vectors which
are perpendicular/orthogonal to each other. The inner product of two mutually

ld
orthogonal vectors is equal to zero. Similarly the inner product of two orthogonal
signals is also equal to zero.
Let ψ k (t) = ej 2πfk t and ψn (t) = ej 2πfn t be two complex exponential signals whose

or
inner product, over the time duration of Ts, is given by:

¸ (i+1)T s
N= ψ k (t).ψ ∗n (t)dt (6.67)
iT s

W
When this integral is evaluated, it is found that if fk and fn are integer multiples
of 1/Ts then N equals zero. This implies that for two harmonics of an exponential
function having a fundamental frequency of 1/Ts, the inner product becomes zero
.But if fk = fn then N equals T s which is nothing but the energy of the complex
exponential signal in the time duration of T s.
TU

Mathematical Description of OFDM

Let us now consider the simultaneous or parallel transmission of P number of com-


plex symbols in the time slot of Ts second (OFDM symbol time duration) and a set
JN

of P orthogonal subcarriers, such that each subcarrier gets amplitude modulated


by a particular symbol from this set of P symbols. Let each orthogonal carrier
. .
be of the form exp j2πn Tt s , where n varies as 0, 1, 2..(P − 1). Here the variable
‘n’ denotes the nth parallel path corresponding to the nth subcarrier. Mathemati-
cally, we can obtain the transmitted signal in Ts seconds by summing up all the P
number of amplitude modulated subcarriers, thereby yielding the following equation:
ll

P −1 . .
. t
p(t) = cngn(t)exp j2πn for 0 ≤ t ≤ Ts (6.68)
A

n=0
T s

125
If p(t) is sampled at t = kTs /P , then the resulting waveform, is:

P −1 . .
. kTs /P
p(k) = cn gn (kTs/P )exp j2πn
Ts
n=0
1 .P −1 . .
k
c n exp j2πn for 0 ≤ k ≤ P − 1 (6.69)
= √ s P
T n=0

ld
This is nothing but the IDFT on the symbol block of P symbols. This can be realized
using IFFT but the constraint is that P has to be a power of 2. So at the receiver,
FFT can be done to get back the required block of symbols. This implementation is

or
better than using multiple oscillators for subcarrier generation which is uneconomical
and since digital technology has greatly advanced over the past few decades, IFFTs
and FFTs can be implemented easily. The frequency spectrum, therefore consists
of a set of P partially overlapping sinc pulses during any time slot of duration T s .

W
This is due to the fact that the Fourier Transform of a rectangular pulse is a sinc
function. The receiver can be visualized as consisting of a bank of demodulators,
translating each subcarrier down to DC, then integrating the resulting signal over a
symbol period to recover the raw data.
TU
But the OFDM symbol structure so generated at the transmitter end needs to
be modified. Since inter symbol interference (ISI) is introduced by the transmission
channel due to multipaths and also due to the fact that when the bandwidth of
OFDM signal is truncated, its effect in the time domain is to cause symbol spreading
such that a part of the symbol overlaps with the adjacent symbols. In order to cope
JN

with ISI as discussed previously the OFDM symbol duration can be increased. But
this might not be feasible from the implementation point of view specifically in terms
of FFT size and Doppler shifts.
A different approach is to keep a guard time interval between two OFDM symbols
in which part of the symbol is copied from the end of the symbol to the front and is
ll

popularly known as the cyclic-prefix. If we denote the guard time interval as Tg and
Ts be the useful symbol duration, then after this cyclical extension the total symbol
duration becomes T = Tg + T s . When the guard interval is longer than the length
A

of the channel impulse response, or the multipath delay, then ISI can be eliminated.
However the disadvantage is the reduction in data rate or throughput and greater
power requirements at the transmitting end. The OFDM transmitter and receiver

126
ld
or
W
Figure 6.13: OFDM Transmitter and Receiver Block Diagram.

sections are as given in the following diagram.


TU
Conclusion

In this chapter, a major chunk has been devoted to digital communication systems
which obviously have certain distinction in comparison to their analog counterpart
due to their signal-space representation. The important modulation techniques for
JN

wireless communication such as QPSK, MSK, GMSK were taken up at length. A


relatively new modulation technology, OFDM, has also been discussed. Certain
practical issues of concern are also discussed. It should be noted that albeit imple-
menting these efficient modulation techniques, the channel still introduces fading in
different ways. In order to prevent that, we need some additional signal processing
techniques mainly at the receiver side. These techniques are discussed in the next
ll

chapter.
A

127
Techniques to Mitigate Fading

ld
Effects

or
Introduction

Apart from the better transmitter and receiver technology, mobile communications
require signal processing techniques that improve the link performance. Equaliza-

W
tion, Diversity and channel coding are channel impairment improvement techniques.
Equalization compensates for Inter Symbol Interference (ISI) created by multipath
within time dispersive channels. An equalizer within a receiver compensates for
the average range of expected channel amplitude and delay characteristics. In other
TU
words, an equalizer is a filter at the mobile receiver whose impulse response is inverse
of the channel impulse response. As such equalizers find their use in frequency selec-
tive fading channels. Diversity is another technique used to compensate fast fading
and is usually implemented using two or more receiving antennas. It is usually em-
ployed to reduce the depths and duration of the fades experienced by a receiver in
JN

a flat fading channel. Channel coding improves mobile communication link perfor-
mance by adding redundant data bits in the transmitted message.At the baseband
portion of the transmitter, a channel coder maps a digital message sequence in to
another specific code sequence containing greater number of bits than original con-
tained in the message. Channel Coding is used to correct deep fading or spectral
ll

null. We discuss all three of these techniques in this chapter. A general framework
of the fading effects and their mitigation techniques is shown in Figure 7.1.
A

128
ld
or
Figure 7.1: A general framework of fading effects and their mitigation techniques.

Equalization

W
ISI has been identified as one of the major obstacles to high speed data transmission
over mobile radio channels. If the modulation bandwidth exceeds the coherence
bandwidth of the radio channel (i.e., frequency selective fading), modulation pulses
TU

are spread in time, causing ISI. An equalizer at the front end of a receiver compen-
sates for the average range of expected channel amplitude and delay characteristics.
As the mobile fading channels are random and time varying, equalizers must track
the time-varying characteristics of the mobile channel and therefore should be time-
JN

varying or adaptive. An adaptive equalizer has two phases of operation: training


and tracking. These are as follows.
Training Mode:

• Initially a known, fixed length training sequence is sent by the transmitter so


that the receiver equalizer may average to a proper setting.
ll

• Training sequence is typically a pseudo-random binary signal or a fixed, of


prescribed bit pattern.
A

• The training sequence is designed to permit an equalizer at the receiver to


acquire the proper filter coefficient in the worst possible channel condition.
An adaptive filter at the receiver thus uses a recursive algorithm to evaluate

129
the channel and estimate filter coefficients to compensate for the channel.

Tracking Mode:

• When the training sequence is finished the filter coefficients are near optimal.

• Immediately following the training sequence, user data is sent.

ld
• When the data of the users are received, the adaptive algorithms of the equal-
izer tracks the changing channel.

• As a result, the adaptive equalizer continuously changes the filter characteris-


tics over time.

or
A Mathematical Framework

The signal received by the equalizer is given by

W
x(t) = d(t) ∗ h (t) + nb (t)

where d(t) is the transmitted signal, h(t) is the combined impulse response of the
transmitter,channel and the RF/IF section of the receiver and nb (t) denotes the
baseband noise.
(7.1)
TU
If the impulse response of the equalizer is heq (t), the output of the equalizer is

ŷ (t) = d (t) ∗ h (t) ∗ heq (t) + nb (t) ∗ heq (t) = d (t) ∗ g (t) + nb (t) ∗ heq (t) . (7.2)

However, the desired output of the equalizer is d(t) which is the original source data.
Assuming nb (t)=0, we can write y(t) = d(t), which in turn stems the following
JN

equation:

g (t) = h (t) ∗ heq (t) = δ (t) (7.3)

The main goal of any equalization process is to satisfy this equation optimally. In
frequency domain it can be written as
ll

Heq (f ) H (f ) = 1 (7.4)

which indicates that an equalizer is actually an inverse filter of the channel. If the
A

channel is frequency selective, the equalizer enhances the frequency components with
small amplitudes and attenuates the strong frequencies in the received frequency

131
spectrum in order to provide a flat, composite received frequency response and
linear phase response. For a time varying channel, the equalizer is designed to track
the channel variations so that the above equation is approximately satisfied.

Zero Forcing Equalization

In a zero forcing equalizer, the equalizer coefficients cn are chosen to force the samples

ld
of the combined channel and equalizer impulse response to zero. When each of the
delay elements provide a time delay equal to the symbol duration T, the frequency
response Heq (f ) of the equalizer is periodic with a period equal to the symbol rate

or
1/T. The combined response of the channel with the equalizer must satisfy Nyquist’s
criterion

Hch (f ) Heq (f ) = 1, |f| < 1/2T (7.5)

frequency response of the channel. W


where Hch (f ) is the folded frequency response of the channel. Thus, an infinite
length zero-forcing ISI equalizer is simply an inverse filter which inverts the folded

Disadvantage: Since Heq (f ) is inverse of Hch (f ) so inverse filter may excessively


amplify the noise at frequencies where the folded channel spectrum has high atten-
TU

uation, so it is rarely used for wireless link except for static channels with high SNR
such as local wired telephone. The usual equalizer model follows a time varying or
adaptive structure which is given next.

A Generic Adaptive Equalizer


JN

The basic structure of an adaptive filter is shown in Figure 7.2. This filter is called
the transversal filter, and in this case has N delay elements, N+1 taps and N+1
tunable complex multipliers, called weights. These weights are updated continuously
by an adaptive algorithm. In the figure the subscript k represents discrete time
index. The adaptive algorithm is controlled by the error signal ek . The error signal
ll

is derived by comparing the output of the equalizer, with some signal dk which is
replica of transmitted signal. The adaptive algorithm uses ek to minimize the cost
A

function and uses the equalizer weights in such a manner that it minimizes the cost
function iteratively. Let us denote the received sequence vector at the receiver and

132
ld
or
W
Figure 7.2: A generic adaptive equalizer.

the input to the equalizer as


TU

T
xk = [xk , xk−1, ....., xk−N ] , (7.6)

and the tap coefficient vector as


wk = [w0 , w1 , ....., wN ]T . (7.7)
k k k
JN

Now, the output sequence of the equalizer yk is the inner product of xk and wk, i.e.,
yk = (x k , w k ) = xT wk = wT x k . (7.8)
k k

The error signal is defined as

ek = dk − yk = dk − xT w k . (7.9)
k
ll

Assuming dk and xk to be jointly stationary, the Mean Square Error (MSE) is given
as
A

MSE = E[e2 k] = E[(dk − yk)2]

= E[(dk − x Tk wk )2]
= E[d 2 ] + w T E[x k x T ]wk − 2E[dk x T ]wk (7.10)
k k k k

133
where wk is assumed to be an array of optimum values and therefore it has been
taken out of the E() operator. The MSE then can be expressed as
MSE = ξ = σ2 + wT Rwk − 2pT wk (7.11)
k k

where the signal variance σ2 = E[d2 ] and the cross correlation vector p between the
d k

desired response and the input signal is defined as

ld
. .
p = E [dk x k ] = E d k xk dk xk−1 dk xk−2 · · · dk xk−N . (7.12)

The input correlation matrix R is defined as an (N + 1) × (N + 1) square matrix,

where

or
xk xk−1 xk xk−2 ··· xk xk−N
x2k
2
. . xk−1xk xk−1 xk−1 xk−2 · · · xk−1 xk−N
R=E xx T
=E x x x x x2 ···x x . (7.13)
k k k−2 k k−2 k−1 k−2 k−2 k−N

xk
.

−N xk
W .
.
xk−N xk−1

Clearly, MSE is a function of wk. On equating


minimum MSE (MMSE) which is known as Wiener solution:
..
.
xk−N xk−2
∂ξ
∂w k
···
··· x2
k−N
...

to 0, we get the condition for


TU
wk = R−1p. (7.14)

Hence, MMSE is given by the equation

MMS E = ξmin = σ2 d− pT w k . (7.15)


JN

Choice of Algorithms for Adaptive Equalization

Since an adaptive equalizer compensates for an unknown and time varying channel,
it requires a specific algorithm to update the equalizer coefficients and track the
channel variations. Factors which determine algorithm’s performance are:
Rate of convergence: Number of iterations required for an algorithm, in re-
ll

sponse to a stationary inputs, to converge close enough to optimal solution. A fast


rate of convergence allows the algorithm to adapt rapidly to a stationary environ-
A

ment of unknown statistics.


Misadjustment: Provides a quantitative measure of the amount by which the
final value of mean square error, averaged over an ensemble of adaptive filters,
deviates from an optimal mean square error.

134
Computational complexity: Number of operations required to make one com-
plete iteration of the algorithm.
Numerical properties: Inaccuracies like round-off noise and representation
errors in the computer, which influence the stability of the algorithm.
Three classic equalizer algorithms are primitive for most of today’s wireless stan-
dards. These include the Zero Forcing Algorithm (ZF), the Least Mean Square Algo-

ld
rithm (LMS), and the Recursive Least Square Algorithm (RLS). Below, we discuss
a few of the adaptive algorithms.

or
Least Mean Square (LMS) Algorithm

LMS algorithm is the simplest algorithm based on minimization of the MSE between
the desired equalizer output and the actual equalizer output, as discussed earlier.
Here the system error, the MSE and the optimal Wiener solution remain the same

W
as given the adaptive equalization framework.
In practice, the minimization of the MSE is carried out recursively, and may be
performed by use of the stochastic gradient algorithm. It is the simplest equalization
algorithm and requires only 2N+1 operations per iteration. The filter weights are
TU
updated by the update equation. Letting the variable n denote the sequence of
iteration, LMS is computed iteratively by

wk (n + 1) = wk (n) + µek (n) x (n − k) (7.16)

where the subscript k denotes the kth delay stage in the equalizer and µ is the step
JN

size which controls the convergence rate and stability of the algorithm.
The LMS equalizer maximizes the signal to distortion ratio at its output within
the constraints of the equalizer filter length. If an input signal has a time dispersion
characteristics that is greater than the propagation delay through the equalizer, then
the equalizer will be unable to reduce distortion. The convergence rate of the LMS
algorithm is slow due to the fact that there is only one parameter, the step size, that
ll

controls the adaptation rate. To prevent the adaptation from becoming unstable,
the value of µ is chosen from
A

,N
.
0<µ <2 λi (7.17)
i=1
where λi is the i-th eigenvalue of the covariance matrix R.

135
Normalized LMS (NLMS) Algorithm

In the LMS algorithm, the correction that is applied to wk (n) is proportional to


the input sample x (n − k). Therefore when x (n − k) is large, the LMS algorithm
experiences gradient noise amplification. With the normalization of the LMS step
size by "x (n)"2 in the NLMS algorithm, this problem is eliminated. Only when
x(n−k) becomes close to zero, the denominator term "x (n)"2 in the NLMS equation

ld
becomes very small and the correction factor may diverge. So, a small positive
number ε is added to the denominator term of the correction factor. Here, the step
size is time varying and is expressed as

or
β
µ (n) = . (7.18)
"x (n)"2 + ε
Therefore, the NLMS algorithm update equation takes the form of

Diversity
wk (n + 1) = wk (n) +

W β
"x (n)" + ε 2
ek (n) x (n − k) . (7.19)

Diversity is a method used to develop information from several signals transmitted


TU

over independent fading paths. It exploits the random nature of radio propagation
by finding independent signal paths for communication. It is a very simple concept
where if one path undergoes a deep fade, another independent path may have a
strong signal. As there is more than one path to select from, both the instantaneous
JN

and average SNRs at the receiver may be improved. Usually diversity decisions are
made by receiver. Unlike equalization, diversity requires no training overhead as a
training sequence is not required by transmitter. Note that if the distance between
two receivers is a multiple of λ/2, there might occur a destructive interference be-
tween the two signals. Hence receivers in diversity technique are used in such a
way that the signal received by one is independent of the other. Diversity can be of
ll

various forms, starting from space diversity to time diversity. We take up the types
one by one in the sequel.
A

136
ld
or
Figure 7.3: Receiver selection diversity, with M receivers.

Different Types of Diversity

Space Diversity
W
A method of transmission or reception, or both, in which the effects of fading are
minimized by the simultaneous use of two or more physically separated antennas,
TU
ideally separated by one half or more wavelengths. Signals received from spatially
separated antennas have uncorrelated envelopes.
Space diversity reception methods can be classified into four categories: selection,
feedback or scanning, maximal ratio combining and equal gain combining.
(a) Selection Diversity:
JN

The basic principle of this type of diversity is selecting the best signal among all
the signals received from different branches at the receiving end. Selection Diversity
is the simplest diversity technique. Figure 7.3 shows a block diagram of this method
where ’M’ demodulators are used to provide M diversity branches whose gains are
adjusted to provide the same average SNR for each branch. The receiver branches
ll

having the highest instantaneous SNR is connected to the demodulator.


Let M independent Rayleigh fading channels are available at a receiver. Each
A

channel is called a diversity branch and let each branch has the same average SNR.
The signal to noise ratio is defined as
Eb
SNR = Γ = α2 (7.20)
N0

137
where Eb is the average carrier energy, N0 is the noise PSD, α is a random variable
used to represent amplitude values of the fading channel.
The instantaneous SNR(γi) is usually defined as γi = instantaneous signal power
per branch/mean noise power per branch. For Rayleigh fading channels, α has a
Rayleigh distribution and so α2 and consequently γi have a chi-square distribution
with two degrees of freedom. The probability density function for such a channel is

ld
1 −γi
Γ . (7.21)
p (γi) = e
Γ
The probability that any single branch has an instantaneous SNR less than some

or
defined threshold γ is
γ γ
¸ ¸ −γi −γ
1
Pr [γi ≤ γ] = p (γi) dγi = e Γ dγi = 1 − e Γ = P (Γ). (7.22)
0 0 Γ

W
Similarly, the probability that all M independent diversity branches receive signals
which are simultaneously less than some specific SNR threshold γ is

Pr [γ 1, γ 2, . . . , γ M ≤ γ] = 1 − e
. −γ
Γ
.M
= PM (γ) (7.23)

where PM (γ) is the probability of all branches failing to achieve an instantaneous


TU

SNR = γ. Quite clearly, PM (Γ) < P (Γ). If a single branch achieves SNR > γ, then
the probability that SNR > γ for one or more branches is given by
. −γ
.M
Pr [γ i > γ] = 1 − P M (γ) = 1 − 1 − e Γ (7.24)
JN

which is more than the required SNR for a single branch receiver. This expression
shows the advantage when a selection diversity is used.
To determine of average signal to noise ratio, we first find out the pdf of γ as
d M. .M −1 e−γ/Γ .
− γ/Γ (7.25)
pM (γ) = PM (γ) = 1 −e
dγ Γ
The average SNR, γ̄, can be then expressed as
ll

¸∞ ¸∞ M−1
. −x.
γ̄ = γpM (γ) dγ = Γ Mx 1 − e e−x dx (7.26)
A

0 0

where x = γ/Γ and Γ is the average SNR for a single branch, when no diversity is
used.

138
This equation shows an average improvement in the link margin without requir-
ing extra transmitter power or complex circuitry, and it is easy to implement as
it needed a monitoring station and an antenna switch at the receiver. It is not an
optimal diversity technique as it doesn’t use all the possible branches simultaneously.
(b) Feedback or Scanning Diversity:
Scanning all the signals in a fixed sequence until the one with SNR more than a

ld
predetermined threshold is identified. Feedback or scanning diversity is very similar
to selection diversity except that instead of always using the best of N signals, the N
signals are scanned in a fixed sequence until one is found to be above a predetermined

or
threshold. This signal is then received until it falls below threshold and the scanning
process is again initiated. The resulting fading statistics are somewhat inferior, but
the advantage is that it is very simple to implement(only one receiver is required).
(c) Maximal Ratio Combining:

W
Signals from all of the m branches are weighted according to their individual
signal voltage to noise power ratios and then summed. Individual signals must be
cophased before being summed, which generally requires an individual receiver and
phasing circuit for each antenna element. Produces an output SNR equal to the
TU
sum of all individual SNR. Advantage of producing an output with an acceptable
SNR even when none of the individual signals are themselves acceptable. Modern
DSP techniques and digital receivers are now making this optimal form, as it gives
the best statistical reduction of fading of any known linear diversity combiner. In
terms of voltage signal,
JN

m
.
rm = ttiri (7.27)
i=1
where tti is the gain and ri is the voltage signal from each branch.
(d) Equal Gain Combining:
In some cases it is not convenient to provide for the variable weighting capability
required for true maximal ratio combining. In such cases, the branch weights are
ll

all set unity, but the signals from each branch are co-phased to provide equal gain
combining diversity. It allows the receiver to exploit signals that are simultaneously
A

received on each branch. Performance of this method is marginally inferior to max-


imal ratio combining and superior to Selection diversity. Assuming all the tti to be

139
ld
or
W
TU

Figure 7.4: Maximal ratio combining technique.

unity, here, m
.
rm = ri . (7.28)
JN

i=1

Polarization Diversity

Polarization Diversity relies on the decorrelation of the two receive ports to achieve
diversity gain. The two receiver ports must remain cross-polarized. Polarization
Diversity at a base station does not require antenna spacing. Polarization diversity
ll

combines pairs of antennas with orthogonal polarizations (i.e. horizontal/vertical, ±


slant 45o, Left-hand/Right-hand CP etc). Reflected signals can undergo polarization
A

changes depending on the channel. Pairing two complementary polarizations, this


scheme can immunize a system from polarization mismatches that would otherwise
cause signal fade. Polarization diversity has prove valuable at radio and mobile com-

140
munication base stations since it is less susceptible to the near random orientations
of transmitting antennas.

Frequency Diversity

In Frequency Diversity, the same information signal is transmitted and received


simultaneously on two or more independent fading carrier frequencies. Rationale

ld
behind this technique is that frequencies separated by more than the coherence
bandwidth of the channel will be uncorrelated and will thus not experience the same
fades. The probability of simultaneous fading will be the product of the individual

or
fading probabilities. This method is employed in microwave LoS links which carry
several channels in a frequency division multiplex mode (FDM). Main disadvantage
is that it requires spare bandwidth also as many receivers as there are channels used
for the frequency diversity.

Time Diversity
W
In time diversity, the signal representing the same information are sent over the
same channel at different times. Time diversity repeatedly transmits information at
TU
time spacings that exceeds the coherence time of the channel. Multiple repetition
of the signal will be received with independent fading conditions, thereby providing
for diversity. A modern implementation of time diversity involves the use of RAKE
receiver for spread spectrum CDMA, where the multipath channel provides redun-
dancy in the transmitted message. Disadvantage is that it requires spare bandwidth
JN

also as many receivers as there are channels used for the frequency diversity. Two
important types of time diversity application is discussed below.

Application 1: RAKE Receiver

In CDMA spread spectrum systems, CDMA spreading codes are designed to provide
ll

very low correlation between successive chips, propagation delay spread in the radio
channel provides multiple version of the transmitted signal at the receiver. Delaying
A

multipath components by more than a chip duration, will appear like uncorrelated
noise at a CDMA receiver. CDMA receiver may combine the time delayed versions
of the original signal to improve the signal to noise ratio at the receiver. RAKE

141
ld
or
Figure 7.5: RAKE receiver.

W
receiver collect the time shifted versions of the original signal by providing a sep-
arate correlation receiver for M strongest multipath components. Outputs of each
correlator are weighted to provide a better estimate of the transmitted signal than
provided by a single component. Demodulation and bit decisions are based on the
TU

weighted output of the correlators. Schematic of a RAKE receiver is shown in Figure


7.5.

Application 2: Interleaver
JN

In the encoded data bits, some source bits are more important than others, and
must be protected from errors. Many speech coder produce several important bits
in succession. Interleaver spread these bit out in time so that if there is a deep fade
or noise burst, the important bits from a block of source data are not corrupted
at the same time. Spreading source bits over time, it becomes possible to make
use of error control coding. Interleaver can be of two forms, a block structure or a
ll

convolutional structure.
A block interleaver formats the encoded data into a rectangular array of m rows
A

and n columns, and interleaves nm bits at a time. Each row contains a word of
source data having n bits. an interleaver of degree m consists of m rows. source bits
are placed into the interleaver by sequentially increasing the row number for each

142
successive bit, and forming the columns. The interleaved source data is then read
out row-wise and transmitted over the channel. This has the effect of separating
the original source bits by m bit periods. At the receiver, de-interleaver stores the
received data by sequentially increasing the row number of each successive bit, and
then clocks out the data row-wise, one word at a time. Convolutional interleavers
are ideally suited for use with convolutional codes.

ld
Channel Coding

or
In channel coding, redundant data bits are added in the transmitted message so
that if an instantaneous fade occurs in the channel, the data may still be recov-
ered at the receiver without the request of retransmission. A channel coder maps
the transmitted message into another specific code sequence containing more bits.

W
Coded message is then modulated for transmission in the wireless channel. Channel
Coding is used by the receiver to detect or correct errors introduced by the channel.
Codes that used to detect errors, are error detection codes. Error correction codes
can detect and correct errors.
TU

Shannon’s Channel Capacity Theorem

In 1948, Shannon showed that by proper encoding of the information, errors induced
by a noise channel can be reduced to any desired level without sacrificing the rate
of information transfer. Shannon’s channel capacity formula is applicable to the
JN

AWGN channel and is given by:

C = B log2 . .
S = B log2
.
P
.
= B log2 . Eb Rb
. (7.29)
1+ N 1 + N0 B 1 + N0 B
where C is the channel capacity (bit/s), B is the channel bandwidth (Hz), P is the
received signal power (W), N0 is the single sided noise power density (W/Hz), Eb is
the average bit energy and Rb is transmission bit rate.
ll

Equation (7.29) can be normalized by the bandwidth B and is given as


C . E R .
= log 1 + b b (7.30)
A

2 N 0 B
B
and the ratio C/B is denoted as bandwidth efficiency. Introduction of redundant
bits increases the transmission bit rate and hence it increases the bandwidth require-
ment, which reduces the bandwidth efficiency of the link in high SNR conditions, but

143
provides excellent BER performance at low SNR values. This leads to the following
two inferences.
Corollary 1 : While dealing within maximum channel capacity, introduction of re-
dundant bits increase the transmitter rate and hence bandwidth requirement also
increases, while decreasing the bandwidth efficiency, but it also decreases the BER.
Corollary 2 : If data redundancy is not introduced in a wideband noisy environment,

ld
error free performance in not possible (for example, CDMA communication in 3G
mobile phones).
A channel coder operates on digital message (or source) data by encoding the source

or
information into a code sequence for transmission through the channel. The error
correction and detection codes are classified into three groups based on their struc-
ture.
1. Block Code
2. Convolution Code
3. Concatenated Code.

Block Codes
W
TU
Block codes are forward error correction (FEC) codes that enable a limited number
of errors to be detected and corrected without retransmission. Block codes can be
used to improve the performance of a communications system when other means of
improvement (such as increasing transmitter power or using a more sophisticated
demodulator) are impractical.
JN

In block codes, parity bits are added to blocks of message bits to make codewords
or code blocks. In a block encoder, k information bits are encoded into n code bits.
A total of n−k redundant bits are added to the k information bits for the purpose of
detecting and correcting errors. The block code is referred to as an (n, k) code, and
the rate of the code is defined as Rc = k/n and is equal to the rate of information
ll

divided by the raw channel rate.


A

Parameters in Block Code

(a) Code Rate (Rc): As defined above, Rc = k/n.


(b) Code Distance (d): Distance between two codewords is the number of ele-

144
ments in which two codewords Ci and Cj differs denoted by d (Ci, Cj ). If the code
used is binary, the distance is known as ’Hamming distance’. For example d(10110,
11011) is 3. If the code ’C’ consists of the set of codewords, then the minimum
distance of the code is given by dmin = min{d (Ci, Cj )}.

(c) Code Weight (w): Weight of a codeword is given by the number of nonzero
elements in the codeword. For a binary code, the weight is basically the number of

ld
1s in the codeword. For example weight of a code 101101 is 4.
Ex 1: The block code C = 00000, 10100, 11110, 11001 can be used to represent two
bit binary numbers as:

or
• 00 – 00000

• 01 – 10100

• 10 – 11110

• 11 – 11001
W
Here number of codewords is 4, k = 2, and n = 5.
To encode a bit stream 1001010011
TU

• First step is to break the sequence in groups of two bits, i.e., 10 01 01 00 11

• Next step is to replace each block by its corresponding codeword, i.e.,


11110 10100 10100 00000 11001

Quite clearly, here, dmin = min{d (Ci, Cj )} = 2.


JN

Properties of Block Codes

(a) Linearity: Suppose Ci and Cj are two code words in an (n, k) block code. Let
α1 and α2 be any two elements selected from the alphabet. Then the code is said to
be linear if and only if α1C1 + α2C2 is also a code word. A linear code must contain
the all-zero code word.
ll

(b) Systematic: A systematic code is one in which the parity bits are appended
to the end of the information bits. For an (n, k) code, the first k bits are identical
A

to the information bits, and the remaining n − k bits of each code word are linear
combinations of the k information bits.

145
(c) Cyclic: Cyclic codes are a subset of the class of linear codes which satisfy the
following cyclic shift property: If C = [Cn−1, Cn−2, ..., C0] is a code word of a cyclic

code, then [Cn−2, Cn−3, ..., C0, Cn−1], obtained by a cyclic shift of the elements of C,
is also a code word. That is, all cyclic shifts of C are code words.
In this context, it is important to know about Finite Field or Galois Field.
Let F be a finite set of elements on which two binary operations – addition (+) and

ld
multiplication (.) are defined. The set F together with the two binary operations is
called a field if the following conditions are satisfied:
1. F is a commutative group under addition.

or
2. The set of nonzero elements in F is a commutative group under multiplication.
3. Multiplication is distributive over addition; that is, for any three elements a, b,
and c in F, a(b + c) = ab + ac
4. Identity elements 0 and 1 must exist in F satisfying a + 0 = a and a.1 = a.

W
5. For any a in F, there exists an additive inverse (−a) such that a + (−a) = 0.
6. For any a in F, there exists an multiplicative inverse a−1 such that a.a−1 = 1.
Depending upon the number of elements in it, a field is called either a finite or an
infinite field. The examples of infinite field include Q (set of all rational numbers),
TU
R (set of all real numbers), C (set of all complex numbers) etc. A field with a finite
number of elements (say q) is called a ’Galois Field’ and is denoted by GF(q). A
finite field entity p(x), called a polynomial, is introduced to map all symbols (with
several bits) to the element of the finite field. A polynomial is a mathematical
expression
JN

p (x) = p0 + p 1 x + ... + pm xm (7.31)

where the symbol x is called the indeterminate and the coefficients p 0, p 1, ..., pm are
the elements of GF(q). The coefficient pm is called the leading coefficient. If pm
is not equal to zero, then m is called the degree of the polynomial, denoted as deg
p(x). A polynomial is called monic if its leading coefficient is unity. The division
ll

algorithm states that for every pair of polynomials a(x) and b(x) in F(x), there
exists a unique pair of polynomials q(x), the quotient, and r(x), the remainder,
A

such that a(x) = q(x)b(x) + r(x), where deg r(x)¡deg b(x). A polynomial p(x) in
F(x) is said to be reducible if p(x)=a(x)b(x), otherwise it is called irreducible. A
monic irreducible polynomial of degree at least one is called a prime polynomial.

146
An irreducible polynomial p(x) of degree ‘m’ is said to be primitive if the smallest
integer ‘n’ for which p(x) divides xn +1 is n = 2 m −1. A typical primitive polynomial
is given by p(x) = xm + x + 1.
A specific type of code which obeys both the cyclic property as well as poly-
nomial operation is cyclic codes. Cyclic codes are a subset of the class of linear
codes which satisfy the cyclic property. These codes possess a considerable amount

ld
of structure which can be exploited. A cyclic code can be generated by using a
generator polynomial g(p) of degree (n-k). The generator polynomial of an (n,k)
cyclic code is a factor of pn + 1 and has the form

or
g (p) = pn−k + gn −k −1pn−k−1 + · · · + g1 p + 1. (7.32)

A message polynomial x(p) can also be defined as

W
x (p) = xk−1pk−1 + · · · + x 1 p + x0

where (xk−1, . . . , x0) represents the k information bits. The resultant codeword c(p)
can be written as
c (p) = x (p) g (p)
(7.33)

(7.34)
TU

where c(p) is a polynomial of degree less than n. We would see an application of


such codes in Reed-Solomon codes.

Examples of Block Codes


JN

(a) Single Parity Check Code: In single parity check codes (example: ASCII code),
an overall single parity check bit is appended to ’k’ information bits. Let the infor-
mation bit word be: (b1, b2, ..., bk), then parity check bit: p = b1 + b2 + ......... + bk
modulo 2 is appended at the (k+1)th position, making the overall codeword: C =
(b1, b2, ..., bk , p). The parity bit may follow an even parity or an odd parity pattern.
All error patterns that change an odd number of bits are detectable, and all even
ll

numbered error patterns are not detectable. However, such codes can only detect
the error, it cannot correct the error.
A

Ex. 2: Consider a (8,7) ASCII code with information codeword (0, 1, 0, 1, 1, 0, 0)


and encoded with overall even parity pattern. Thus the overall codeword is (0, 1, 0,
1, 1, 0, 0, 1) where the last bit is the parity bit. If there is a single error in bit 3: (0,

147
1, 1, 1, 1, 0, 0, 1), then it can be easily checked by the receiver that now there are
odd number of 1’s in the codeword and hence there is an error. On the other hand,
if there are two errors, say, errors in bit 3 and 5: (0, 1, 1, 1, 0, 0, 0, 1), then error
will not be detected.
After decoding a received codeword, let pc be the probability that the decoder
gives correct codeword C, pe is the probability that the decoder gives incorrect

ld
codeword Ct ƒ= C, and pf is the probability that the decoder fails to give a codeword.
In this case, we can write pc + pe + pf = 1.
If in an n-bit codeword, there are j errors and p is the bit error probability,
then the probability of obtaining j errors in this codeword is Pj = n C j pj (1 − p)n−j .

or
Using this formula, for any (n, n − 1) single parity check block code, we get

• pc = P0,

• pe = P2 + P4 + ... + P tn (nt = n if n is even, otherwise nt = n − 1),

W
• pf = P1 + P3 + ... + P tn (nt = n − 1 if n is even, otherwise nt = n).

As an example, for a (5,4) single parity check block code, pc = P0, pe = P2 + P4,
and pf = P1 + P3 + P5.
TU
(b) Product Codes: Product codes are a class of linear block codes which pro-
vide error detection capability using product of two block codes. Consider that nine
information bits (1, 0, 1, 0, 0, 1, 1, 1, 0) are to be transmitted. These 9 bits can be
divided into groups of three information bits and (4,3) single parity check codeword
can be formed with even parity. After forming three codewords, those can be ap-
JN

pended with a vertical parity bit which will form the fourth codeword. Thus the
following codewords are transmitted:

C1 = [1 0 1 0]

C2 = [0 0 1 1]

C3 = [1 1 0 0]
ll

C4 = [0 1 0 1].
A

Now if an error occurs in the second bit of the second codeword, the received code-
words at the receiver would then be

C1 = [1 0 1 0]

148
C2 = [0 1 1 1] ←

C3 = [1 1 0 0]

C4 = [0 1 0 1]

and these would indicate the corresponding row and column position of the erroneous

ld
bit with vertical and horizontal parity check. Thus the bit can be corrected. Here
we get a horizontal (4, 3) codeword and a vertical (4, 3) codeword and concatenating
them we get a (16, 9) product code. In general, a product code can be formed as

or
(n1, k1) & (n2, k2) → (n1n2, k1k2).
(c) Repetition Codes: In a (n,1) repetition code each information bit is repeated
n times (n should be odd) and transmitted. At the receiver, the majority decoding
principle is used to obtain the information bit. Accordingly, if in a group of n received

W
bit, 1 occurs a higher number of times than 0, the information bit is decoded as 1.
Such majority scheme works properly only if the noise affects less than n/2 number
of bits. Ex 3: Consider a (3,1) binary repetition code.

• For input bit 0, the codeword is (0 0 0) and for input bit 1, the codeword is
TU
(1 1 1).

• If the received codeword is (0 0 0), i.e. no error, it is decoded as 0.

• Similarly, if the received codeword is (1 1 1), i.e. no error, it is decoded as 1.

• If the received codeword is (0 0 1) or (0 1 0) or (1 0 0), then error is detected


and it is decoded as 0 with majority decoding principle.
JN

• If the received codeword is (0 1 1) or (1 1 0) or (1 0 1), once again error is


detected and it is decoded as 1 with majority decoding principle.

For such a (3,1) repetition code, pc = P0 + P1, pe = P2 + P 3, and pf = 0.


(d) Hamming Codes: A binary Hamming code has the property that

(n, k) = (2m − 1, 2m − 1 − m)
ll

(7.35)

where k is the number of information bits used to form a n bit codeword, and m
A

is any positive integer. The number of parity symbols are n − k = m. Thus, a

149
codeword is represented by C = [i1, ...in, p1, ..., pn−k]. This is quite a useful code in
communication which is illustrated via the following example.

Ex 4: Consider a (7, 4) Hamming code. With three parity bits we can correct exactly
1 error. The parity bits may follow such a modulo 2 arithmetic:

ld
p 1 = i1 + i2 + i 3 ,

p 2 = i2 + i3 + i 4 ,
p3 = i1 + i3 + i 4 ,

or
which is same as,

p 1 + i1 + i2 + i3 = 0

p 2 + i2 + i3 + i4 = 0

W
p3 + i1 + i3 + i4 = 0.

The transmitted codeword is then C = [i1, i2, ..., i 4, p1, p2, p3].
Syndrome Decoding: For this Hamming code, let the received codeword be V =

[v1, v2, ..., v4, v5, v6, v7]. We define a syndrome vector S as
TU

S = [S1 S2 S3]

S1 = v1 + v2 + v3 + v5

S2 = v2 + v3 + v4 + v6

S3 = v1 + v2 + v4 + v7
JN

It is obvious that in case of no error, the syndrome vector is equal to zero. Corre-
sponding to this syndrome vector, there is an error vector e which can be obtained
from a syndrome table and finally the required codeword is taken as C = V + e. In
a nutshell, to obtain the required codeword, we perform the following steps:
ll

1. Calculate S from decoder input V.


2. From syndrome table, obtain e corresponding to S.
3. The required codeword is then C = V + e.
A

A few cases are given below to illustrate the syndrome decoding.


1. Let C = [0 1 1 1 0 1 0] and V = [0 1 1 1 0 1 0]. This implies S = [0 0 0], and it
corresponds to e = [0 0 0 0 0 0 0]. Thus, C = V + e = [0 1 1 1 0 1 0].

150
2. Let C = [1 1 0 0 0 1 0] and V = [1 1 0 1 0 1 0]. This means S = [0 1 1], from which
we get e = [0 0 0 1 0 0 0] which means a single bit error is there in the received bit

v4. This will be corrected by performing the operation C = V + e.


3. Another interesting case is, let C = [0 1 0 1 1 0 0] and V = [0 0 1 1 1 0 1] (two
errors at second and third bits). This makes S = [0 0 0] and as a result, e = [0 0 0 0 0
0 0]. However, C ƒ= V , and C = V + e implies the double error cannot be

ld
corrected. Therefore a (7,4) Hamming code can correct only single bit error.
(e) Golay Codes: Golay codes are linear binary (23,12) codes with a minimum
distance of seven and a error correction capability of three bits. This is a special,

or
one of a kind code in that this is the only nontrivial example of a perfect code.
Every codeword lies within distance three of any codeword, thus making maximum
likelihood decoding possible.
(f) BCH Codes: BCH code is one of the most powerful known class of linear

W
cyclic block codes, known for their multiple error correcting ability, and the ease
of encoding and decoding. It’s block length is n = 2m − 1 for m ≥ 3 and number
of errors that they can correct is bounded by t < (2m − 1)/2. Binary BCH codes
can be generalized to create classes of non binary codes which use m bits per code
TU
symbol.
(g) Reed Solomon (RS) Codes: Reed-Solomon code is an important subset of
the BCH codes with a wide range of applications in digital communication and data
storage. Typical application areas are storage devices (CD, DVD etc.), wireless
communications, digital TV, high speed modems. It’s coding system is based on
JN

groups of bits, such as bytes, rather than individual 0 and 1. This feature makes it
particularly good at dealing with burst of errors: six consecutive bit errors. Block
length of these codes is n = 2m − 1, and can be extended to 2m or 2 m + 1. Number
of parity symbols that must be used to correct e errors is n − k = 2e. Minimum
distance dmin = 2e + 1, and it achieves the largest possible d min of any linear code.
ll

For US-CDPD, the RS code is used with m = 6. So each of the 64 field elements
is represented by a 6 bit symbol. For this case, we get the primitive polynomial as
A

p(x) = x6 + x + 1. Equating p(x) to 0 implies x6 = x + 1.


The 6 bit representation of the finite field elements is given in Table 7.1. The table
elements continue up to α62. However, to follow linearity property there should be

151
Table 7.1: Finite field elements for US-CDPD
α5 α4 α3 α2 α1 α0
1 0 0 0 0 0 1
α1 0 0 0 1 0 0
α2 0 0 1 0 0 0

ld
. . . . . . .
. . . . . . .
α6 = α + 1 0 0 0 0 1 1
. . . . . . .

or
. . . . . . .

a zero codeword, hence α63 is assigned zero.

W
The encoding part of the RS polynomial is done as follows:
Information polynomial: d(x) = Cn−1xn−1 + Cn−2x n−2 + ..... + C2tx 2t ,
Parity polynomial: p(x) = C2t−1x2t−1 + ... + C 0,
Codeword polynomial: c(x) = d(x) + p(x).
Since generating an information polynomial is difficult, so a generating polynomial is
TU
used instead. Information polynomial is then the multiple of generating polynomial.
This process is given below.
Since this kind of codes are cyclic codes, we take a generating polynomial g(x)
such that d(x) = g(x)q(x) + r(x) where q(x) is the quotient polynomial and r(x)
is the remainder polynomial. The codeword polynomial would then be given as:
JN

c(x) = g(x)q(x) + r(x) = p(x). If we assign a parity polynomial p(x) = r(x), then
the codeword polynomial c(x) = g(x)p(x) and the entire process becomes easier.
On the decoder side one has to find a specific r(x) = p(x) or vice-versa, but due
to its complexity, it is mainly done using syndrome calculation. The details of such
a syndrome calculation can be found in [1].
ll

Convolutional Codes
A

A continuous sequence of information bits is mapped into a continuous sequence of


encoder output bits. A convolutional code is generated by passing the information
sequence through a finite state shift register. Shift register contains ’N’ k-bit stages

152
ld
Figure 7.6: A convolutional encoder with n=2 and k=1.

or
and m linear algebraic function generators based on the generator polynomials.
Input data is shifted into and along the shift register, k-bits at a time. Number
of output bits for each k-bit user input data sequence is n bits, so the code rate
Rc = k/n. The shift register of the encoder is initialized to all-zero-state before

W
TU
JN

Figure 7.7: State diagram representation of a convolutional encoder.

encoding operation starts. It is easy to verify that encoded sequence is 00 11 10


00 01 . . . for an input message sequence of 01011 . . .. Convolution codes may be
represented in various ways as given below.
ll

State Diagram:
Since the output of the encoder is determined by the input and the current
state of the encoder, a state diagram can be used to represent the encoding process.
A

The state diagram is simply a graph of the possible states of the encoder and the
possible transitions from one state to another. The path information between the

states, denoted as b/c1c2, represents input information bit ’b’ and the corresponding

153
ld
or
Figure 7.8: Tree diagram representation of a convolutional encoder.

W
output bits (c1c2). Again, it is not difficult to verify from the state diagram that an
input information sequence b = (1011) generates an encoded sequence c = (11, 10,
00, 01).
Tree Diagram:
TU
The tree diagram shows the structure of the encoder in the form of a tree with the
branches representing the various states and the outputs of the coder. The encoded
bits are labeled on the branches of the tree. Given an input sequence, the encoded
sequence can be directly read from the tree. As an example, an input sequence
(1011) results in the encoded sequence (11, 10, 00, 01).
JN
ll
A

Figure 7.9: Trellis diagram of a convolutional encoder.

154
ld
Figure 7.10: Block diagram of a turbo encoder.

or
Trellis Diagram:
Tree reveals that the structure repeats itself once the number of stages is greater
than the constraint length. It is observed that all branches emanating from two

W
nodes having the same state are identical in the sense that they generate identical
output sequences. This means that the two nodes having the same label can be
merged. By doing this throughout the tree diagram, we obtain another diagram
called a Trellis Diagram which is more compact representation.
TU

Concatenated Codes

Concatenated codes are basically concatenation of block and convolutional codes. It


can be of two types: serial and parallel codes. Below, we discuss a popular parallel
concatenated code, namely, turbo code.
JN

Turbo Codes: A turbo encoder is built using two identical convolutional codes
of special type with parallel concatenation. An individual encoder is termed a com-
ponent encoder. An interleaver separates the two component encoders. The inter-
leaver is a device that permutes the data sequence in some predetermined manner.
Only one of the systematic outputs from the two component encoders is used to form
ll

a codeword, as the systematic output from the other component encoder is only a
permuted version of the chosen systematic output. Figure 7.10 shows the block di-
A

agram of a turbo encoder using two identical encoders. The first encoder outputs

the systematic V0 and recursive convolutional V1 sequences while the second en-
coder discards its systematic sequence and only outputs the recursive convolutional

V2 sequence. Depending on the number of input bits to a component encoder it

155
may be binary or m-binary encoder. Encoders are also categorized as systematic
or non-systematic. If the component encoders are not identical then it is called an
asymmetric turbo code.

ld
or
W
TU
JN
ll
A

156
ld
Equalization, Diversity, and Channel Coding

or
•Introduction

W
•Equalization Techniques
•Algorithms for Adaptive Equalization
•Diversity Techniques

TU
•RAKE Receiver
•Channel Coding
JN
ll
A

NCCU
7-1 Wireless Comm. Lab.
ld
Introduction[1]

or
•Three techniques are used independently or in tandem to improve

W
receiver signal quality
•Equalization compensates for ISI created by multipath with time
dispersive channels (W>BC)

TU
¾Linear equalization, nonlinear equalization
•Diversity also compensates for fading channel impairments, and is
usually implemented by using two or more receiving antennas
JN
¾Spatial diversity, antenna polarization diversity, frequency
diversity, time diversity
ll
A

NCCU
7-2 Wireless Comm. Lab.
ld
Introduction[1]

or
•The former counters the effects of time dispersion (ISI), while the
latter reduces the depth and duration of the fades experienced

W
by a receiver in a flat fading (narrowband) channel
• Channel Coding improves mobile communication link
performance by adding redundant data bits in the transmitted

TU
message
•Channel coding is used by the Rx to detect or correct some (or all)
of the errors introduced by the channel (Post detection
JN
technique)
¾Block code and convolutional code
ll
A

NCCU
7-3 Wireless Comm. Lab.
ld
Equalization Techniques

or
• The term equalization can be used to describe any signal

W
processing operation that minimizes ISI [2]
• Two operation modes for an adaptive equalizer: training
and tracking

TU
•Three factors affect the time spanning over which an
equalizer converges: equalizer algorithm, equalizer
structure and time rate of change of the multipath radio
JN
channel
•TDMA wireless systems are particularly well suited for
equalizers
ll
A

NCCU
7-4 Wireless Comm. Lab.
ld
Equalization Techniques

or
• Equalizer is usually implemented at baseband or at IF in a
receiver (see Fig. 1)

W
y( t ) = x( t ) ∗ f ∗( t ) + n ( t )
b

TU
f*(t): complex conjugate of f(t)
nb(t): baseband noise at the input of the equalizer
heq(t): impulse response of the equalizer
JN
ll
A

NCCU
7-5 Wireless Comm. Lab.
ld
Equalization Techniques

or
W
TU
JN

Fig. 1
ll
A

NCCU
7-6 Wireless Comm. Lab.
ld
Equalization Technologies

or
dˆ (t ) = y (t ) ∗ heq (t )
= x (t ) ∗ f ∗ (t ) ∗ heq (t ) + mb (t ) ∗ heq (t )

W
= δ(t )
∴ F ∗ (− f ) ∗ H eq ( f ) = 1

TU
• If the channel is frequency selective, the equalizer enhances the
frequency components with small amplitudes and attenuates the strong
JN
frequencies in the received frequency response
• For a time-varying channel, an adaptive equalizer is needed to track the
channel variations
ll
A

NCCU
7-7 Wireless Comm. Lab.
ld
Basic Structure of Adaptive Equalizer

or
•Transversal filter with N delay elements, N+1 taps, and N+1 tunable
complex weights

W
TU
JN

•These weights are updated continuously by an adaptive algorithm


ll

•The adaptive algorithm is controlled by the error signal ek


A

NCCU
7-8 Wireless Comm. Lab.
ld
Equalization Techniques

or
•Classical equalization theory : using training sequence to minimize
the cost function

W
E[e(k) e*(k)]
•Recent techniques for adaptive algorithm : blind algorithms
¾Constant Modulus Algorithm (CMA, used for constant envelope

TU
modulation) [3]
¾Spectral Coherence Restoral Algorithm (SCORE, exploits spectral
JN
redundancy or cyclostationarity in the Tx signal) [4]
ll
A

NCCU
7-9 Wireless Comm. Lab.
ld
Solutions for Optimum Weights of Figure 2 (一)

or
•Error signal e k = x k − y kT ω k = x k − ω kT y k
where y k = [ y k y k −1 y k − 2 .... y k − N ]T

W
ω k = [ωk ωk − 1 ωk − 2 .... ωk − N ]T

= x k2 + ω y k y kT ω − 2 x k y kT ω
2
•Mean square error ek T
k k k

TU
•Expected MSE ξ = E ek [ ] = E [x ]+ ω
2 2
k
T
Rω − 2pTω
where ⎡ yk2 yk yk −1 yk yk − 2 .... yk yk − N ⎤
⎢ ⎥
[ ]y y
R = E yk yk = E ⎢ k −1 k
yk2−1 yk −1 yk −2 .... yk −1 yk − N ⎥
JN
*
⎢ .... .... .... .... .... ⎥
⎢ 2 ⎥
⎣ yk − N yk yk − N yk −1 yk − N yk − 2 .... yk − N ⎦

p = E[ x k y k ] = E[ x k y k xk yk −2 .... xk yk − N ]
T
xk yk −1
ll
A

NCCU
7-10 Wireless Comm. Lab.
ld
Solutions for Optimum Weights of Figure 2 (二)

or
•Optimum weight vector
ˆ = R −1p
ω

W
•Minimum mean square error (MMSE)
[ ] T −1
ξmin = E χ κ2 − p R p
= E[χ ] −

TU
2
κ
pΤω̂
•Minimizing the MSE tends to reduce the bit error rate
JN
ll
A

NCCU
7-11 Wireless Comm. Lab.
ld
Equalization Techniques

or
•Two general categories - linear and nonlinear
equalization (see Fig. 3)

W
•In Fig. 1, if d(t) is not the feedback path to adapt the equalizer,
the equalization is linear
•In Fig. 1, if d(t) is fed back to change the subsequent outputs

TU
of the equalizer, the equalization is nonlinear
JN
ll
A

NCCU
7-12 Wireless Comm. Lab.
ld
Equalization Techniques

or
W
TU
JN
ll

Fig.3 Classification of equalizers


A

NCCU
7-13 Wireless Comm. Lab.
ld
Equalizer Techniques

or
•Linear transversal equalizer (LTE, made up of tapped delay lines
as shown in Fig.4)

W
TU
JN
Fig.4 Basic linear transversal equalizer structure

•Finite impulse response (FIR) filter (see Fig.5)


•Infinite impulse response (IIR) filter (see Fig.5)
ll
A

NCCU
7-14 Wireless Comm. Lab.
ld
Equalizer Techniques

or
W
TU
JN

Fig.5 Tapped delay line filter with both feedforward and feedback taps
ll
A

NCCU
7-15 Wireless Comm. Lab.
ld
Structure of a Linear Transversal Equalizer [5]

or
W
TU
N2
• dˆ k = ∑ C n* y k − n
n=− N1

[ ]
JN
T πT No
• E e(n) = ∫ dω
2
π
2π T F( e ) + N o
− jωt 2

F(e jωt ) :frequency response of the channel


ll

N o :noise spectral density


A

NCCU
7-16 Wireless Comm. Lab.
ld
Structure of a Lattice Equalizer [6-7]

or
W
TU
JN

Fig.7 The structure of a Lattice Equalizer


ll
A

NCCU
7-17 Wireless Comm. Lab.
ld
Characteristics of Lattice Filter

or
•Advantages
¾Numerical stability

W
¾Faster convergence
¾Unique structure allows the dynamic assignment of the most effective
length

TU
•Disadvantages
¾The structure is more complicated
JN
ll
A

NCCU
7-18 Wireless Comm. Lab.
ld
Nonlinear Equalization

or
•Used in applications where the channel distrotion is too severe
•Three effective methods [6]

W
¾Decision Feedback Equalization (DFE)
¾Maximum Likelihood Symbol Detection
¾Maximum Likelihood Sequence Estimator (MLSE)

TU
JN
ll
A

NCCU
7-19 Wireless Comm. Lab.
ld
Nonlinear Equalization--DFE

or
•Basic idea : once an information symbol has been detected and decided
upon, the ISI that it induces on future symbols can be estimated and

W
substracted out before detection of subsequent symbols
•Can be realized in either the direct transversal form (see Fig.8) or as a
lattice filter

TU
N2 N3
• dˆ k = ∑ C y k − n + ∑ F i d k − i
*
n
n=− N1 i=1
JN
[
• E e(n)
2
] T Tπ
= exp{ ∫− π ln[
2π T
No
]dω}
F( e ) + N o
jωT 2
min
ll
A

NCCU
7-20 Wireless Comm. Lab.
ld
Nonlinear Equalizer-DFE

or
W
TU
JN

Fig.8 Decision feedback equalizer (DFE)


ll
A

NCCU
7-21 Wireless Comm. Lab.
ld
Nonlinear Equalization--DFE

or
•Predictive DFE (proposed by Belfiore and Park, [8])
•Consists of an FFF and an FBF, the latter is called a noise predictor

W
( see Fig.9 )
•Predictive DFE performs as well as conventional DFE as the limit
in the number of taps in FFF and the FBF approach infinity

TU
•The FBF in predictive DFE can also be realized as a lattice structure [9].
The RLS algorithm can be used to yield fast convergence
JN
ll
A

NCCU
7-22 Wireless Comm. Lab.
ld
Nonlinear Equalizer-DFE

or
W
TU
JN

Fig.9 Predictive decision feedback equalizer


ll
A

NCCU
7-23 Wireless Comm. Lab.
ld
Nonlinear Equalization--MLSE

or
•MLSE tests all possible data sequences (rather than decoding each
received symbol by itself ), and chooses the data sequence with the

W
maximum probability as the output
•Usually has a large computational requirement
•First proposed by Forney [10] using a basic MLSE estimator

TU
structure and implementing it with the Viterbi algorithm
•The block diagram of MLSE receiver (see Fig.10 )
JN
ll
A

NCCU
7-24 Wireless Comm. Lab.
ld
Nonlinear Equalizer-MLSE

or
W
TU
Fig.10 The structure of a maximum likelihood sequence equalizer(MLSE) with
JN
an adaptive matched filter
•MLSE requires knowledge of the channel characteristics in order
to compute the matrics for making decisions
•MLSE also requires knowledge of the statistical distribution of
ll

the noise corrupting the signal


A

NCCU
7-25 Wireless Comm. Lab.
ld
Algorithm for Adaptive Equalization

or
•Excellent references [6, 11--12]
•Performance measures for an algorithm

W
¾Rate of convergence
¾Misadjustment
¾Computational complexity

TU
¾Numerical properties
•Factors dominate the choice of an equalization structure and its algorithm
¾The cost of computing platform
JN
¾The power budget
¾The radio propagation characteristics
ll
A

NCCU
7-26 Wireless Comm. Lab.
ld
Algorithm for Adaptive Equalization

or
•The speed of the mobile unit determines the channel fading rate and the
Dopper spread, which is related to the coherent time of the channel

W
directly
•The choice of algorithm, and its corresponding rate of convergence,
depends on the channel data rate and coherent time

TU
•The number of taps used in the equalizer design depends on the maximum
expected time delay spread of the channel
•The circuit complexity and processing time increases with the number of
JN
taps and delay elements
ll
A

NCCU
7-27 Wireless Comm. Lab.
ld
Algorithm for Adaptive Equalization

or
•Three classic equalizer algorithms : zero forcing (ZF), least mean squares
(LMS), and recursive least squares (RLS) algorithms

W
•Summary of algorithms (see Table 1)

TU
JN
ll
A

NCCU
7-28 Wireless Comm. Lab.
ld
Summary of algorithms

or
W
TU
JN

Table 1 Comparison of various algorithms for adaptive equalization


ll
A

NCCU
7-29 Wireless Comm. Lab.
ld
Diversity Techniques

or
•Requires no training overhead
•Can provides significant link improvement with little added cost

W
•Diversity decisions are made by the Rx, and are unknown to the Tx
•Diversity concept
¾If one radio path undergoes a deep fade, another independent path may

TU
have a strong signal
¾By having more than one path to select from, both the instantaneous
and average SNRs at the receiver may be improved, often by as much
JN
as 20 dB to 30 dB
ll
A

NCCU
7-30 Wireless Comm. Lab.
ld
Diversity Techniques

or
•Microscopic diversity and Macroscopic diversity
¾The former is used for small-scale fading while the latter for large-scale

W
fading
¾Antenna diversity (or space diversity)
•Performance for M branch selection diversity (see Fig.11)

TU
Pr[SNR > r ] = 1 − Pr [γ 1 , .... , γ M ≤ r ]
= 1 − (1 − e − r/Γ )M
Μ
JN
PM (r) = Pr[SNR≤ r] = (1− e−r/Γ )M−1e−r/Γ
d
dr Γ
M
r 1
Γ
= ∑
ll

k =1 k
A

NCCU
7-31 Wireless Comm. Lab.
ld
Diversity techniques

or
W
TU
JN

Fig. 11 Graph of probability distributions of SNR=γ threshold for M branch


ll

selection diversity. The term Γ represents the mean SNR on each branch
A

NCCU
7-32 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Performance for Maximal Ratio Combining Diversity [13]
(see Fig. 12)

W
M
γ M = ∑ Giγ i
M
N T = N ∑ Gi
2

i =1 i =1

TU
γ M2
rM =
2 NT
( r / Γ ) k −1
JN
r M
Pr{rM ≤ r} = ∫ p ( rM )drM = 1 − e −r / Γ

0 k =1 ( k − 1)!

M −1 − rM / Γ
rM
e
P ( rM ) = M
Γ ( M − 1)!
ll
A

NCCU
7-33 Wireless Comm. Lab.
ld
Diversity Techniques

or
W
TU
JN
Fig. 12 Generalized block diagram for space diversity
ll
A

NCCU
7-34 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Space diversity [14]

W
¾ Selection diversity

TU
¾ Feedback diversity
JN
¾ Maximal ration combining

¾ Equal gain diversity


ll
A

NCCU
7-35 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Selection diversity (see Fig. 13)
¾ The receiver branch having the highest instantaneous SNR

W
is connected to the demodulator
¾The antenna signals themselves could be sampled and the

TU
best one sent to a single demodulation
JN
ll

Fig. 13 Maximal ratio combiner


A

NCCU
7-36 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Feedback or scanning diversity (see Fig. 14)
¾ The signal, the best of M signals, is received until it falls

W
below threshold and the scanning process is again initiated

TU
JN
ll

Fig. 14 Basic form for scanning diversity


A

NCCU
7-37 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Maximal ratio combining [15] (see Fig. 12)

W
¾ The signals from all of the M branches are weighted
according to their signal voltage to noise power ratios and
then summed

TU
• Equal gain diversity

¾ The branch weights are all set to unity but the signals from
JN
each are co-phased to provide equal gain combining
diversity
ll
A

NCCU
7-38 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Polarization diversity
¾ Theoretical model for polarization diversity [16] (see Fig.15)

W
x = r1 cos(ωt + φ1 )
the signal arrive at the base station
y = r2 cos(ωt + φ2 )
the correlation coefficient can be written as

TU
2
⎛ tan 2 (α ) cos2 ( β ) − Γ ⎞
ρ = ⎜⎜ 2 ⎟⎟
⎝ tan (α ) cos 2
(β ) + Γ ⎠
JN
R22
Γ =
R12

R1 = r12a2 + r22b2 + 2r1r2ab cos(φ1 + φ2 )


ll

R1 = r12a2 + r22b2 − 2r1r2ab cos(φ1 + φ2 )


A

NCCU
7-39 Wireless Comm. Lab.
ld
Diversity Techniques

or
W
TU
JN
ll

Fig. 15 Theoretical Model for base station polarization diversity based on [Koz85]
A

NCCU
7-40 Wireless Comm. Lab.
ld
Diversity Techniques

or
• Frequency diversity
¾ Frequency diversity transmits information on more than one

W
carrier frequency
¾ Frequencies separated by more than the coherence bandwidth

TU
of the channel will not experience the same fads
• Time diversity
JN
¾ Time diversity repeatedly transmits information at time
spacings that exceed the coherence time of the channel
ll
A

NCCU
7-41 Wireless Comm. Lab.
ld
RAKE Receiver

or
• RAKE Receiver [17]
M Z m2
Z′ = ∑αm Z m αm =

W
M
m =1 ∑ Z m2
m =1

TU
JN

Fig. 16 An M-branch (M-finger) RAKE receiver implementation. Each correlator detects a time shifted
version of the original CDMA transmission, and each finger of the RAKE correlates to a portion of the
ll

signal which is delayed by at least one chip in time from the other finger.
A

NCCU
7-42 Wireless Comm. Lab.
ld
Interleaving

or
W
TU
JN

Fig. 17 Block interleaver where source bits are read into columns and out as n-bit rows
ll
A

NCCU
7-43 Wireless Comm. Lab.
ld
References

or
[1] T. S. Rappaport, Wireless Communications -- Principles and Practice, Prentice Hall Inc., New Jersey, 1996.
[2] S.U.H. Qureshi, “Adaptive equalization, ” Proceeding of IEEE, vol. 37 no.9, pp.1340 -- 1387, Sept. 1985.
[3] J. R. Treichler, and B.G. Agoe, “A new approach to multipath correction of constant modulus signals, ”

W
IEEE Trans. Acoustics, Speech, and Signal Processing, vol. ASSP--31, pp. 459--471, 1983
[4] W. A. Gardner, “Exploitation of spectral redundancy in cyclostationary signals, ” IEEE Signal Processing
Magazine, pp. 14-- 36, April 1991.
[5] I.Korn, Digital Communications, Van Nostrand Reinhold, 1985.

TU
[6] J. Proakis, “Adaptive equalization for TDMA digital mobile radio, ” IEEE Trans. Commun., vol. 40, no.2,
pp.333--341, May 1991.
[7] J. A. C. Bingham, The Theory and Practice of Modem Design, John Wiley & sons, New York.
[8] C. A, Belfiori, and J.H. Park, “Decision feedback equalization, ” Proceedings of IEEE, vol. 67, pp. 1143--
JN
1156, Aug. 1979.
[9] K. Zhou, J.G. Proakis, F. Ling, “Decision feedback equalization of time dispersive channels with coded
modulation, ” IEEE Trans. Commun., vol. 38, pp. 18--24, Jan. 1990.
ll
A

NCCU
7-44 Wireless Comm. Lab.
ld
References

or
[10] G. D. Forney, “The Viterbi algorithm, ” Proceedings of the IEEE, vol.61, no.3, pp. 268--278, March 1978.
[11] B. Widrow, and S.D. Stearns, Adaptive Signal Processing, Prentice Hall, 1985.

W
[12] S. Haykin, Adaptive Filter Theory, Prentice Hall, Englewood Cliffs, NJ, 1986.
[13] T. Eng, N. Kong, and L. B. Milstein, “Comparison of Diversity Combining Techniques for Rayleigh-
Fading Channels,” IEEE Trans. Commun., vol. 44, pp. 1117-1129, Sep. 1996.
[14] W. C. Jakes, “A Comparision of Space Diversity Techniques for Reduction of Fast Fading in UHF Mobile
Radio Systems,” IEEE Trans. Veh. Technol., vol. VT-20, No. 4, pp. 81-93,

TU
Nov. 1971.
[15] L. Kahn, “Radio Square,” Proceedings of IRE, vol. 42, pp. 1074, Nov. 1954.
[16] S. Kozono, et al, “Base Station Polarization Diversity Reception for Mobile Radio,” IEEE Trans. Veh.
Technol., vol VT-33, No. 4, pp. 301-306, Nov. 1985.
JN
[17] R. Price, P. E. Green, “A Communication Technique for Multipath Channel,” Proceeding of the IRE, pp.
555-570, March 1958
ll
A

NCCU
7-45 Wireless Comm. Lab.
ld
Chapter 7:
Equalization and

or
Diversity

W
School of Information Science
TU
and Engineering, SDU
J N
ll
A
ld
Outline

or
l Introduction
l Fundamentals of Equalization

W
l Survey of Equalization Techniques
l Linear Equalizers
l Nonlinear Equalization

l
l
l

TU
Algorithms for Adaptive Equalization
Fundamentals of diversity
Survey of Diversity Techniques
N
l Frequency/Time/Space/Polarization Diversity
l Selection/MRC/EGC Combining
J

l RAKE Receiver
l Interleaving
ll
A
ld
7.1 Introduction

or
l The properties of mobile radio channels:
l Multipath fading -> time dispersion, ISI

W
l Doppler spread -> dynamical fluctuation
These effects have a strong negative impact on the bit error rate of any
modulation.
l

l
TU
Mobile communication systems require signal processing
techniques that improve the link performance in hostile
mobile radio environments.
Three popular techniques:
N
l Equalization: compensates for ISI
J

l Diversity: compensates for channel fading


l Channel coding: detects or corrects errors
ll

These techniques can be deployed independently or jointly.


A
ld
Transmitted signal: s (t )
h(t ) = ∑ α k δ (t − τ k )

or
Channel model:
k

W
Received signal: r (t ) = s (t ) ∗ h(t ) = ∑ α k s (t − τ k )
k
Ts Ts

TU α1
N
α3
α4
J

α2
ll
A
ld
(1) Equalization

or
l If the modulation bandwidth exceeds the coherence bandwidth
of the radio channel, ISI occurs and modulation pulses are

W
spread in time.

l Equalization compensates for intersymbol interference (ISI)


created by multipath within time dispersive channels.

TU
An equalizer within a receiver compensates for the average range
of expected channel amplitude and delay characteristics.
N
l Equalizers must be adaptive
since the channel is generally unknown and time varying.
J
ll
A
ld
(2) Diversity

or
l Usually employed to reduce the depth and duration of the
fades experienced by a receiver in a flat fading (narrowband)

W
channel.
Without increasing the transmitted power or bandwidth.
l Can be employed at both base station and mobile receivers.
l

TU
Types of diversity:
.antenna polarization diversity
.frequency diversity
N
.time diversity.
For example, CDMA systems often use a RAKE receiver,
J

which provides link improvement through time diversity


l Spatial diversity is the most common one.
ll

While one antenna sees a signal null, one of the other


antennas may see a signal peak.
A
ld
(3) Channel Coding

or
l Used to Improve mobile communication link performance by
adding redundant data bits in the transmitted message.

W
At the baseband portion of the transmitter, a channel coder
maps a digital message sequence into another specific code
sequence containing a greater number of bits than originally
contained in the message.

l
wireless channel.TU
The coded message is then modulated for transmission in the
N
l coding can be considered to be a post detection technique.
Because decoding is performed after the demodulation portion
J

l two general types of channel codes:


ll

block codes
convolutional codes.
A
ld
or
l Channel coding is generally treated independently from the
type of modulation used

W
but this has changed recently with the use of trellis coded
modulation schemes that combine coding and modulation to
achieve large coding gains without any bandwidth expansion.

Notes
l
TU
The three techniques of equalization, diversity, and channel
N
coding are used to improve radio link performance (i.e. to
minimize the instantaneous bit error rate)
J

l but the approach, cost, complexity, and effectiveness of each


technique varies widely in practical wireless communication
ll

systems.
A
7.2 Fundamentals of Equalization

ld
or
l Intersymbol interference (ISI)

W
l caused by multipath propagation (time dispersion) ;
l cause bit errors at the receiver;
l the major obstacle to high speed data transmission over

TU
mobile radio channels.

l Equalization
N
l a technique used to combat ISI;
l can be any signal processing operation that minimizes ISI;
J

l usually track the varying channel adaptively.


ll
A
ld
Operating modes of an adaptive equalizer

or
l Training (first stage)
l A known fixed-length training sequence is sent by the
transmitter so that the receiver's equalizer may average to a

W
proper setting.
l The training sequence is designed to permit an equalizer at
the receiver to acquire the proper filter coefficients in the
worst possible channel conditions

TU
The training sequence is typically a pseudorandom binary signal
or a fixed, prescribed bit pattern.
Immediately following the training sequence, the user data is
sent.
N
l The time span over which an equalizer converges is a
function of
J

1. the equalizer algorithm


2. the equalizer structure
ll

3. the time rate of change of the multipath radio channel.


Equalizers require periodic retraining in order to maintain
A

effective ISI cancellation.


ld
Operating modes of an adaptive equalizer

or
l Tracking (second stage)
Immediately following the training sequence, the user data is sent.

W
l As user data are received, the adaptive algorithm of the
equalizer tracks the changing channel and adjusts its filter
characteristics over time.

l
TU
commonly used in digital communication systems
where user data is segmented into short time blocks.
N
l TDMA wireless systems are particularly well suited for
equalizers.
J

data in fixed-length time blocks,


training sequence usually sent at the beginning of a block
ll
A
ld
Communication system with an adaptive equalizer

or
l Equalizer can be implemented at baseband or at IF in a receiver.
l Since the baseband complex envelope expression can be used

W
to represent bandpass waveforms and, thus, the channel
response, demodulated signal, and adaptive equalizer
algorithms are usually simulated and implemented at baseband

TU
Block diagram of a simplified communications system using an
adaptive equalizer at the receiver is shown in next page
J N
ll
A
ld
Communication system with an adaptive equalizer

or
Radio
x(t )

W
Modulator Transmitter RF Front End
Channel

Detector
nb (t ) + IF Stage

TU
Matched Filter
f (t )
y (t )
Adaptive Decision
N
heq (t ) Equalizer Maker
d (t )

dˆ (t )
J

Σ
ll

e(t )
A
ld
Relevant equations

or
y (t ) = x(t ) ∗ f (t ) + nb (t )
dˆ (t ) = x(t ) ∗ f (t ) ∗ heq (t ) + nb (t ) ∗ heq (t )

W
heq (t ) = ∑ ck δ (t − nTs )

TU
k

To eliminate ISI, we must have

heq (t ) ∗ f (t ) = δ (t ) H eq ( f ) =
1
N
F( f )
J

an equalizer is an inverse filter of the channel.


In frequency selective channel, enhances the frequency
components with small amplitudes, attenuates the strong
ll

frequencies
A

therefore provide a flat, composite, received frequency response


and linear phase response.
7.3 A Generic Adaptive Equalizer

ld
or
yk yk-1 yk-2

W
Z-1 Z-1 Z-1 Z-1

w0 w1
TU w2 wN
dˆk
Σ
N
Adaptive algorithm that updates the weights

ek
J

Σ
ll

Prior knowledge: d k
A
7.3 A Generic Adaptive Equalizer

ld
or
l A transversal filter with
l N delay elements

W
l N+1 taps

l N+1 tunable complex multipliers

l N+1 weights:

l
algorithm TU
These weights are updated continuously by the adaptive

either on a sample by sample basis or on a block by block basis.


N
l The adaptive algorithm is controlled by the error signal ek.
J

ek is derived by comparing the output of the equalizer with some


signal which is either an exact scaled replica of the transmitted
signal xk or which represents a known property of the transmitted
ll

signal.
A
ld
7.3 A Generic Adaptive Equalizer

or
l A cost function is used
the cost function is minimized by using ekThe, and the weights

W
are updated iteratively.
l For example, The least mean squares (LMS) algorithm can
serve as a cost function.

TU
l Iterative operation based on LMS
New weights = Previous weights + (constant) x (Previous error) x (Current input vector)
Where
Previous error = Previous desired output — Previous actual output
N
This process is repeated rapidly in a programming loop while the
equalizer attempts to converge
J

Upon reaching convergence, the adaptive algorithm freezes the


filter weights until the error signal exceeds an acceptable level or
ll

until a new training sequence is sent.


A
ld
7.3 A Generic Adaptive Equalizer

or
l Techniques used to minimize the error
l gradient

W
l steepest decent algorithms

l Based on classical equalization theory, the most common cost


function is MSE
TU
MSE----mean square error (MSE) between the desired signal
and the output of the equalizer
N
Denoted by E[e(k ) ⋅ e (k )]
*
J
ll
A
ld
7.3 A Generic Adaptive Equalizer

or
Blind algorithms
l more recent class of adaptive algorithms

W
l able to exploit characteristics of the transmitted signal and do
not require training sequences.
provide equalizer convergence without burdening the
transmitter with training overhead

TU
able to acquire equalization through property restoral
techniques of the transmitted signal,
l Two techniques:
l the constant modulus algorithm (CMA)
N
used for constant envelope modulation
forces the equalizer weights to maintain a constant envelope on
J

the received signal


l spectral coherence restoral algorithm (SCORE).
ll

exploits spectral redundancy or cyclostationarity in the


transmitted signal
A
7.4 Equalizers in a Communications

ld
Receiver

or
l Because noise is present, an equalizer is unable to achieve
perfect performance.

W
l Therefore, the instantaneous combined frequency response
will not always be flat, resulting in some finite prediction error.
l The mean squared error (MSE) E [ek2] is one of the most

l
TU
important measures of how well an equalizer works.
Minimizing MSE E [ek2] tends to reduce the bit error rate.
For wireless communication links, it would be best to minimize
N
the instantaneous probability of error instead of MSE
generally results in nonlinear equations
J

much more difficult to solve in real-time


ll
A
ld
7.5 Survey of Equalization Techniques

or
l Equalization techniques can be subdivided into two general
categories:

W
l linear equalization
l The output of the decision maker is not used in the feedback
path to adapt the equalizer.
l nonlinear equalization

l
TU
l The output of the decision maker is used in the feedback path
to adapt the equalizer.

Many filter structures are used to implement linear and


N
nonlinear equalizers
J

l For each structure, there are numerous algorithms used to


adapt the equalizer.
ll
A
ld
Classification of equalizers

or
Equalizer

W
Linear Nonlinear

Types

TU
ML Symbol
DFE MLSE
Detector
N
Transversal
Structures Transversal Lattice Transversal Lattice
Channel Est.
J

Zero forcing Gradient RLS LMS Gradient RLS LMS


LMS RLS
ll

RLS
Algorithms RLS Fast RLS Fast RLS
A

Fast RLS Sq. root RLS Sq. root RLS


Sq. root RLS
ld
Most common structure:
---- Linear transversal equalizer (LTE)

or
l made up of tapped delay lines, with the tappings spaced a
symbol period (Ts) apart

W
l the transfer function can be written as a function of the delay
operator − jωTs or Z −1
Assuming that the delay elements have unity gain and delay Ts,

TU
of a linear N
J
ll
A

Basic linear transversal equalizer structure


ld
Most common structure:
---- Linear transversal equalizer (LTE)

or
Two types of LTE

W
l finite impulse response (FIR) filter
l The simplest LTE uses only feedforward Z −1 taps
l Transfer function is a polynomial in
l has many zeroes but poles only at z = 0

l
TU
Usually simply called a transversal filter

Infinite impulse response (IIR) filter


N
l has both feedforward and feedback taps
l transfer function is a rational function of Z-1 with poles
J

and zeros.
l tend to be unstable when used in channels where the
strongest pulse arrives after an echo pulse (i.e., leading
ll

echoes)
rarely used.
A
ld
or
W
TU
J N
ll

Tapped delay line filter with both feedforward and feedback taps (IIR)
A
ld
7.6 Linear Equalizers

or
Transversal filter implementation (LTE)

W
Input

TU
J N
Output
ll

Threshold Detector
A

This type of equalizer is the simplest.


ld
7.6 Linear Equalizers

or
l current and past values of the received signal are linearly
weighted by the filter coefficient and summed to produce the

W
output,
If the delays and the tap gains are analog, the continuous output
of the equalizer is sampled at the symbol rate and the samples are
applied to the decision device.

TU
Implementation is usually carried out in the digital domain where
the samples of the received signal are stored in a shift register.

l The output before decision making (threshold detection)


J N

l The minimum MSE it can achieve


ll
A
ld
7.6 Linear Equalizers

or
Lattice filter implementation

W
TU
J N
ll
A

Numerical stable, faster convergence, Complicated


ld
7.6 Linear Equalizers

or
l Two main advantages of the lattice equalizer
l numerical stability

W
l faster convergence

l Unique structure allows dynamic assignment of the most


effective length
TU
l When channel is not very time dispersive

Only a fraction of the stages are used.


N
l channel becomes more time dispersive
J

Length can be increased without stopping the operation


ll

l Drawback: more complicated than LTE


A
ld
7.7 Nonlinear Equalization

or
l Linear equalizers do not perform well on channels which have
deep spectral nulls in the passband.

W
In an attempt to compensate for the distortion, the linear
equalizer places too much gain in the vicinity of the spectral null,
thereby enhancing the noise present in those frequencies.

TU
Nonlinear equalizers are used in applications where the
channel distortion is too severe for a linear equalizer to
handle.
N
l Three very effective nonlinear equalizer
l Decision Feedback Equalization (DFE)
J

l Maximum Likelihood Symbol Detection


l Maximum Likelihood Sequence Estimation (MLSE)
ll
A
7.7.1 Decision Feedback Equalization

ld
(DFE)

or
Basic idea:
once an information symbol has been detected, the ISI

W
that it induces on future symbols can be estimated and
subtracted out before detection of subsequent symbols.

l DFE Can be realized in either the direct transversal form or

l
TU
as a lattice filter.

The LTE form consists of a feedforward filter (FFF) and a


feedback filter (FBF).
N
The FBF is driven by decisions on the output of the detector,
and its coefficients can be adjusted to cancel the ISI on the
J

current symbol from past detected symbols.


ll

l The equalizer has N1 + N2 + I taps in FFF and N3 taps in FBF


A
7.7.1 Decision Feedback Equalization

ld
(DFE)

or
Input

W
TU
Feedforward Filter
Output
J N
ll
A

Feedback Filter
7.7.1 Decision Feedback Equalization

ld
(DFE)

or
The output of DFE

W
The minimum mean square error of DFE

TU
• It can be seen that the minimum MSE for a DFE is always
smaller than that of an LTE
N
Unless F (e jωT ) is a constant, where adaptive equalization is not
needed
J

• If there are nulls in the F (e jωT ) , a DFE has significantly


smaller minimum MSE than an LTE.
ll
A
7.7.1 Decision Feedback Equalization

ld
(DFE)

or
Conclusion

W
l an LTE is well behaved when the channel spectrum is
comparatively flat
l a DFE is more appropriate for severely distorted wireless
channels.

l
spectrum
l
TU
If the channel is severely distorted or exhibits nulls in the

the performance of an LTE deteriorates and the mean


N
squared error of a DFE is much better than a LTE.
l Also, an LTE has difficulty equalizing a nonminimum phase
J

channel
where the strongest energy arrives after the first arriving signal
component.
ll
A
ld
Another form of DFE----predictive DFE

or
l also consists of a feed forward filter (FFF) as in the
conventional DFE.

W
l Difference: the feedback filter (FBF) is driven by an input
sequence formed by the difference of the output of the
detector and the output of the feed forward filter.

TU
the FBF here is called a noise predictor because it predicts
the noise and the residual ISI contained in the signal at the FFF
output and subtracts from it
N
l The predictive DFE performs as well as the conventional
DFE as the limit in the number of taps in the FFF and the
J

FBF approach infinity.


ll

l The FEF in the predictive DFE can also be realized as a


lattice structure
A
ld
Another form of DFE----predictive DFE

or
W
TU
J N
ll
A
7.7.2 Maximum Likelihood Sequence

ld
Estimation (MLSE) equalizer

or
The MSE-based linear equalizers are optimum with respect to the
criterion of minimum probability of symbol error when the channel

W
does not introduce any amplitude distortion.
Yet this is precisely the condition in which an equalizer is needed
for a mobile communications link.

l
receiver structure.
TU
MLSE uses various forms of the classical maximum likelihood

the MLSE tests all possible data sequences (rather than


N
decoding each received symbol by itself), and chooses the data
sequence with the maximum probability as the output.
J

A channel impulse response simulator is used within the


algorithm,
ll

l Drawback: An MLSE usually has a large computational


requirement
A

especially when the delay spread of the channel is large.


7.7.2 Maximum Likelihood Sequence

ld
Estimation (MLSE) equalizer

or
W
Matched
Filter

TU
Delay
N
Channel
Estimator
J
ll
A
7.7.2 Maximum Likelihood Sequence

ld
Estimation (MLSE) equalizer

or
l The MLSE can be viewed as a problem in estimating the state
of a discrete time finite state machine

W
The channel has ML states, where M is the size of the symbol
alphabet of the modulation.

l An ML trellis is used by the receiver to model the channel over


time.

TU
The Viterbi algorithm then tracks the state of the channel by the
paths through the trellis.
N
l The MLSE is optimal in the sense that it minimizes the
probability of a sequence error.
J
ll
A
7.7.2 Maximum Likelihood Sequence

ld
Estimation (MLSE) equalizer

or
NOTES:

W
l The MLSE requires knowledge of the channel characteristics
in order to compute the metrics for making decisions.

The MLSE also requires knowledge of the statistical

TU
l
distribution of the noise corrupting the signal
the probability distribution of the noise determines the form of the
metric for optimum demodulation of the received signal.
N
l The matched filter operates on the continuous time signal,
whereas the MLSE and channel estimator rely on discretized
J

(nonlinear) samples.
ll
A
ld
7.8 Algorithms for Adaptive Equalization

or
l Equalizer requires a specific algorithm to update the coefficients
and track the channel variations.

W
Since it compensates for an unknown and time-varying channel

l This section outlines three of the basic algorithms for adaptive


equalization.
TU
Though the algorithms detailed in this section are derived for the
linear, transversal equalizer, they can be extended to other equalizer
N
structures, including nonlinear equalizers.
J
ll
A
ld
7.8 Algorithms for Adaptive Equalization

or
Factors determining the performance of an algorithm:

W
l Rate of convergence (fast or slow?)
l Defined as the number of iterations required for the algorithm,
in response to stationary inputs, to converge close enough to
the optimum solution.
l

l
TU
A fast rate of convergence allows the algorithm to adapt
rapidly to a stationary environment of unknown statistics.
Furthermore, it enables the algorithm to track statistical
variations when operating in a nonstationary environment.
N
l Misadjustment (precise or not?)
J

l Provides a quantitative measure of the amount by which the


final value of the mean square error, averaged over an
ll

ensemble of adaptive filters, deviates from the optimal


minimum mean square error.
A
ld
7.8 Algorithms for Adaptive Equalization

or
Factors determining the performance of an algorithm:

W
l Computational complexity (simple or complex?)
l Number of operations required to make one complete
iteration of the algorithm.

l
l TU
Numerical properties (stable or not?)
When an algorithm is implemented numerically, inaccuracies
are produced due to round-off noise and representation
N
errors in the computer.
l These kinds of errors influence the stability of the algorithm.
J
ll
A
ld
7.8 Algorithms for Adaptive Equalization

or
Practical considerations for choice of an equalizer structure and
its algorithm

W
l The cost of the computing platform (affordable or not?)
especially when used in user equipments

l
TU
The power budget (power limited applications or else?)
In portable radio applications, battery drain at the subscriber unit
is a paramount consideration
N
l The radio propagation characteristics (fast fading & time
delay spread?)
J

The speed of the mobile unit determines the channel fading rate
and the Doppler spread, which is directly related to the coherence
ll

time of the channel


A
ld
7.8 Algorithms for Adaptive Equalization

or
Three classic equalizer algorithms

W
l Zero Forcing Algorithm (ZF)
l Least Mean Square Algorithm (LMS)

TU
l Recursive Least Squares Algorithm (RLS)
N
Please read references for detailed
J

information on a specific algorithm.


ll
A
ld
7.8.1 Zero Forcing (ZF) Algorithm

or
Criterion:
to force the samples of the combined channel and equalizer

W
impulse response to zero at all but one of sample points in the
tapped delay line filter.

Disadvantage:
TU
may excessively amplify noise at frequencies where the
folded channel spectrum has high attenuation.
N
Suitability:
J

Wireline communications
1 1
ll

H eq ( f ) = , f <
H ch ( f ) 2T
A
7.8.2 Least Mean Square (LMS)

ld
Algorithm

or
Criterion:

W
to minimize the mean square error (MSE) between the desired
equalizer output and the actual equalizer output.
Minimize ξ = E[ek ⋅ ek ]
*

TU
Must be solved iteratively
Simplest algorithm, requires only 2N + I operations per iteration.

l The LMS equalizer maximizes the signal to distortion ratio at its


N
output within the constraints of the equalizer filter length.

α
J

l a step size is used to control the convergence rate and the


stability
ll
A
7.8.2 Least Mean Square (LMS)

ld
Algorithm

or
Disadvantage: low convergence rate.
Because of the only one parameter α

W
Especially when the eigenvalues of the input covariance matrix
RNN have a very large spread, i.e, λmax / λmin >> 1

l If an input signal has a time dispersion characteristic that is

l
TU
greater than the propagation delay through the equalizer, then
the equalizer will be unable to reduce distortion.

To prevent the adaptation from becoming unstable, the value of α


N
is chosen from
J

where λi is the ith eigenvalue of the covariance matrix RNN.

The step size α can be controlled by the total input power in


ll

l
order to avoid instability in the equalizer [Hay86].
A

since
7.8.3 Recursive Least Squares (RLS)

ld
Algorithm

or
l RLS is Proposed to improve the convergence rate of LMS algorithm.

W
l Error measures expressed in terms of a time average of the actual
received signal instead of a statistical average.

l
the equalizer.
TU
λ is the weighting coefficient that can change the performance of

If a channel is time-invariant, λ can be set to 1. Usually 0.8-1.


N
l
l The value of λ has no influence on the rate of convergence, but
does determine the tracking ability.
J

l The smaller the λ , the better the tracking ability of the equalizer.
l However, if λ is too small, the equalizer will be unstable
ll
A
7.8.3 Recursive Least Squares (RLS)

ld
Algorithm

or
Advantage: high convergence rate

W
Disadvantage: sometimes unstable

TU
The RLS algorithm described above, called the
Kalman RLS algorithm
N
Uses 2.5N2 + 4.5N arithmetic operations per
iteration.
J
ll
A
ld
7.8.4 Summary of equalization algorithms

or
l There are number of variations of the LMS and RLS algorithms
l RLS algorithms have similar convergence and tracking

W
performances, which are much better than the LMS algorithm.
Usually have high computational requirement and complex
program structures.
Some RLS algorithms tend to be unstable.

About FTF
TU
l Among the RLS algorithms, fast transversal filter (FTF)
algorithm requires the least computation
N
l a rescue variable can be used to avoid instability.
However, rescue techniques tend to be a bit tricky for widely
J

varying mobile radio channels.


FTF is not widely used.
ll
A
Comparison of Various Algorithms for Adaptive Equalization
[Pro9l]

ld
or
W
TU
J N
ll
A
ld
7.9 Fractionally Spaced Equalizers(FSE)

or
l In the presence of channel distortion, the matched filter prior to
the equalizer must be matched to the channel and the corrupted

W
signal.
Usually get the suboptimal result because the channel response is
unknown.
This results in a significant degradation in performance.

l
the Nyquist rate. TU
FSE is based on sampling the incoming signal at least as fast as
N
l The FSE compensates for the channel distortion before aliasing
effects occur due to the symbol rate sampling.
J

FSE incorporates the functions of a matched filter and equalizer


into a single filter structure.
ll

l Simulation results demonstrate the effectiveness of the FSE over


A

a symbol rate equalizer. (see [Ste94])


7.10 Fundamentals of Diversity

ld
Techniques

or
l Random nature of radio propagation:
l Multipath propagation

l Independent fading of each Multipath component

W
l If one radio path undergoes a deep fade, another
independent path may have a strong signal

l
TU
Diversity exploits the random nature of radio propagation by
finding independent signal paths for communication, so as to
boost the instantaneous SNR at the receiver.
N
Path 2
J

Transmitter Path 1 Receiver


ll

Path 3
A
7.10 Fundamentals of Diversity

ld
Techniques

or
l Diversity is a powerful communication receiver technique that
provides wireless link improvement at relatively low cost.

W
l Requires no training

l In virtually all applications, diversity decisions are made by the

TU
receiver, and are unknown to the transmitter.

Two types of diversity


N
l Microscopic diversity small scale fading
l Macroscopic diversity large scale fading
J
ll
A
7.10 Fundamentals of Diversity

ld
Techniques

or
Microscopic diversity
l Small-scale fades: deep and rapid amplitude fluctuations over
distances of just a few wavelengths.

W
caused by multiple reflections from the surroundings in the vicinity
of the mobile.
results in a Rayleigh fading distribution of signal strength over small
distances.

l
changing signal.
TU
Microscopic diversity techniques can exploit the rapidly
N
For example, use two antennas at the receiver (separated by a
fraction of a meter), one may receive a null while the other receives a
strong signal.
J

By selecting the best signal at all times, a receiver can mitigate


small-scale fading effects
ll

Called antenna diversity or space diversity


A

l Samples: Rake receiver, MIMO transmission


7.10 Fundamentals of Diversity

ld
Techniques

or
Macroscopic diversity

W
l Large-scale fading: caused by shadowing due to variations in
both the terrain profile and the nature of the surroundings.
In deeply shadowed conditions, the received signal strength at a
mobile can drop well below that of free space.

TU
log-normally distributed with a standard deviation of about 10 dB in
urban environments.

l Macro-scope diversity: By selecting a base station which is not


N
shadowed when others are, the mobile can improve substantially
the average ratio on the forward link.
It is the mobile that takes advantage of large separations between
J

the serving base stations.


ll
A
7.10 Fundamentals of Diversity

ld
Techniques

or
Macroscopic diversity

W
l Macroscopic diversity is also useful at the base station receiver.
By using base station antennas that are sufficiently separated in
space, the base station is able to improve the reverse link by selecting
the antenna with the strongest signal from the mobile.

l
TU
Used to combat slow fading (shadowing)

Samples: Base-station handoff in cellular networks


J N
ll
A
7.10 Fundamentals of Diversity

ld
Techniques

or
Macro-scope diversity

W
Base station Base station

TU
N
Mobile
J
ll
A
7.10 Fundamentals of Diversity

ld
Techniques

or
l Strategies used in diversity techniques
l Selection diversity

W
l Maximal ratio combining diversity

l Equal-gain combining diversity

l Hybrid schemes

l TU
Practical considerations
l effectiveness, complexity, cost, and etc.
J N
ll
A
7.10.1 Derivation of Selection

ld
Diversity improvement

or
l Consider M independent Rayleigh fading channels available
areceiver.

W
Each channel is called a diversity branch.

TU
J N
ll
A
7.10.1 Derivation of Selection

ld
Diversity improvement

or
l Further assume that each branch has the same average SNR
given by

W
Where we assume α = 1 .
2

l If each branch has an instantaneous SNR = γ i , then the pdf of γ i


is
TU
where Γ is the mean SNR of each branch.
N
l The probability that a single branch has SNR less than some
threshold y γ is
J
ll
A
7.10.1 Derivation of Selection

ld
Diversity improvement

or
l Now, the probability that all M independent diversity branches
receive signals which are simultaneously less than some

W
specific SNR threshold γ is

l TU
This is the probability of all branches failing to achieve SNR = γ i .
If a single branch achieves SNR > γ , then the probability that
SNR > γ for one or more branches is given by
J N
This is the probability of exceeding a threshold when selection
diversity is used.
ll
A
A
ll
JN
TU
W
or
ld
7.10.1 Derivation of Selection

ld
Diversity improvement

or
How to determine the average signal-to-noise ratio of the received
signal when diversity is used?

W
l First of all, find the pdf of γ (the instantaneous SNR when M
branches are used). Thus we compute the derivation of CDF
PM (γ ) ,

l TU
Then, we can compute the average SNR, γ ,
N
where x = γ / Γ .
J

The above equation can be evaluated to yield the average


ll

SNR improvement offered by selection diversity.


γ M
1
=∑
A

Γ k =1 k
7.10.1 Derivation of Selection

ld
Diversity improvement

or
l Selection diversity offers an average improvement in the link
margin without requiring additional transmitter power or
sophisticated receiver circuitry.

W
The diversity improvement can be directly related to the average
bit error rate for various modulations.

TU
Selection diversity is easy to implement because all that is
needed is a side monitoring station and an antenna switch at
the receiver.
N
l However, it is not an optimal diversity technique because it
J

does not use all of the possible branchessimultaneously.


Maximal ratio combining uses each of the M branches in a co-
ll

phased and weighted manner such that the highest achievable SNR
is available at the receiver at all times.
A
7.10.1 Derivation of Selection

ld
Diversity improvement

or
Example
Assume four branch diversity is used, where each branch
receives an independent Rayleigh fading signal. If the average

W
SNR is 20 dB, determine the probability that the SNR will drop
below 10 dB. Compare this with the case of a single receiver
without diversity.
Solution
TU
J N
ll
A
7.10.2 Derivation of Maximal Ratio

ld
Combining Improvement

or
In maximal ratio combining, the voltage signals ri from each of
the M diversity branches are co-phased to provide coherent voltage

W
addition and are individually weighted to provide optimal SNR.

TU
J N
ll
A
7.10.2 Derivation of Maximal Ratio

ld
Combining Improvement

or
1) The SNR out of the diversity combiner:
l If each branch has gain Gi , then the resulting signal envelope

W
applied to the detector is

l
TU
Assuming that each branch has the same average noise power
N, the total noise power NT applied to the detector is simply the
weighted sum of the noise in each branch. Thus
J N

which results in an SNR applied to the detector, γ M , given by


ll
A
7.10.2 Derivation of Maximal Ratio

ld
Combining Improvement

or
l Using Chebychev's inequality, γ M is maximized
when Gi = ri / N , which leads to

W
(7-66)

TU
l Conclusion:
The SNR out of the diversity combiner is simply the sum of
the SNRs in each branch.
J N
ll
A
7.10.2 Derivation of Maximal Ratio

ld
Combining Improvement

or
2) The pdf of γ M

W
l According to Chapter 3, γ M is a Chi-square distribution of 2M
Gaussian random variables. Thus, the pdf for γ M is

TU
(7-68)

3) The CDF of γ M
N
l γM
According to the abovementioned pdf, The probability that
is less than some SNR threshold γ is
J
ll
A
7.10.2 Derivation of Maximal Ratio

ld
Combining Improvement

or
γM
4) The average SNR out of the diversity combiner,

W
l γ M can be calculated by using the pdf of γ M (Eq. (7.68)). But the
direct way is to calculate it from Eq. (7-66).

l
TU
That is to say, the average SNR, γ M , is simply the sum of the
individual γ i from each branch.
N
The control algorithms for setting the gains and phases for
J

maximal ratio combining receivers are similar to those required in


equalizers and RAKE receivers.
ll

Maximal ratio combining can be applied to virtually any diversity


application, although often at much greater cost and complexity than
A

other diversity techniques.


7.10.3 Practical Space Diversity

ld
Considerations

or
l Space diversity (also known as antenna diversity), is one of the
most popular forms of diversity used in wireless systems.

W
l The signals received from spatially separated antennas on the
mobile would have essentially uncorrelated envelopes for
antenna separations of one half wavelength or more.

l
or both.
TU
Space diversity can be used at either the mobile or base station,
N
Since the important scatterers are generally on the ground in the
J

vicinity of the mobile, when base station diversity is used, the


antennas must be spaced considerably far apart to achieve
ll

decorrelation (several tens of wavelengths).


A
7.10.3 Practical Space Diversity

ld
Considerations

or
general block diagram of a space diversity scheme

W
TU
J N
ll
A
7.10.3 Practical Space Diversity

ld
Considerations

or
Space diversity reception methods can be classified into four
categories

W
l 1. Selection diversity
l 2. Feedback diversity

TU
l 3. Maximal ratio combining
l 4. Equal gain diversity
J N
ll
A
7.10.3 Practical Space Diversity

ld
Considerations

or
(1) Selection Diversity

W
l The simplest diversity technique.
l The receiver branch having the highest instantaneous SNR is
connected to the demodulator.
l

l
TU
The antenna signals themselves could be sampled and the best
one sent to a single demodulator.
In practice, the branch with the largest (S + N) /N is used, since
it is difficult to measure SNR.
N
l A practical selection diversity system cannot function on a
truly instantaneous basis, but must be designed so that the
J

internal time constants of the selection circuitry are shorter


than the reciprocal of the signal fading rate.
ll
A
7.10.3 Practical Space Diversity

ld
Considerations

or
(2) Feedback or Scanning Diversity
l Very similar to selection diversity

W
l The M signals are scanned in a fixed sequence until one is
found to be above a predetermined threshold.
l This signal is then received until it falls below threshold and

TU
the scanning process is again initiated.
l The resulting fading
statistics are
somewhat inferior to
N
those obtained by
the other methods.
J

l Advantage: very
simple to implement
ll

(only one receiver is


required).
A
7.10.3 Practical Space Diversity

ld
Considerations

or
(3) Maximal Ratio Combining
l The signals from all of the M branches are weighted and then

W
summed.
l The individual signals must be co-phased before being
summed.

TU
requires an individual receiver and phasing circuit for each
antenna element.
l Output SNR equal to the sum of the individual SNRs.
l Advantage: produces an output with an acceptable SNR even
N
when none of the individual signals are themselves acceptable.
l Gives the best statistical reduction of fading of any known
J

linear diversity combiner.


ll
A
7.10.3 Practical Space Diversity

ld
Considerations

or
Maximal Ratio Combiner

W
TU
J N
ll
A
7.10.3 Practical Space Diversity

ld
Considerations

or
(4) Equal Gain Combining

W
In certain cases, it is not convenient to provide for the variable
weighting capability required for true maximal ratio combining. In
such cases, the branch

l
TU
Equal gain combining diversity sets all weights to unity but the
signals from each branch are co-phased.
The possibility of producing an acceptable signal from a
N
number of unacceptable inputs is still retained,
l The performance is only marginally inferior to maximal ratio
J

combining and superior to selection diversity.


ll
A
ld
7.10. 4 PolarIzation Diversity

or
At the base station, space diversity is considerably less practical .

W
l polarization diversity only provides two diversity branches, but
allows the antenna elements to be co-located.
l Measured horizontal and vertical polarization paths between a

TU
mobile and a base station are reported to be uncorrelated.
l Decorrelation for the signals in each polarization is caused by
multiple reflections.
l The reflection coefficient for each polarization is different,
N
which results in different amplitudes and phases for each, or at
least some, of the reflections.
J

l After sufficient random reflections, the polarization state of the


signal will be independent of the transmitted polarization.
ll

In practice, however, there is some dependence of the received


polarization on the transmitted polarization.
A
ld
7.10.5 Frequency Diversity

or
l Transmits information on more than one carrier frequency.
frequencies separated by more than the coherence bandwidth of

W
the channel will not experience the same fades.

l Frequency diversity is often employed in microwave LOS links.

l
TU
In practice, 1:N protection switching is provided by a radio licensee,
When diversity is needed, the appropriate traffic is simply switched
to the backup frequency.
N
Disadvantage: not only requires spare bandwidth but also
J

l
requires that there be as many receivers as there are channels
used for the frequency diversity.
ll

for critical traffic, the expense may be justified.


A
ld
7.10.5 Frequency Diversity

or
l New OFDM modulation and access techniques exploit
frequency diversity by providing simultaneous modulation

W
signals with error control coding across a large bandwidth.

l If a particular frequency undergoes a fade, the composite


signal will still be demodulated.

TU
J N
ll
A
ld
7.10.6 Time Diversity

or
l Time diversity repeatedly transmits information at time
spacings that exceed the coherence time of the channel

W
Multiple repetitions of the signal will be received with independent
fading conditions.

TU
l One modem implementation of time diversity involves the use
of the RAKE receiver for spread spectrum CDMA, where the
multipath channel provides redundancy in the transmitted
message.
J N
ll
A
ld
7.11 RAKE Receiver

or
l In CDMA spread spectrum systems, the spreading codes are
designed to provide very low correlation between successive

W
chips.

l If the multipath components are delayed in time by more than a


chip duration, they appear like uncorrelated noise at a CDMA

TU
receiver, and equalization is not required.

l However, since there is useful information in the multipath


components, CDMA receivers may combine the time delayed
N
versions of the original signal transmission in order to improve
the signal to noise ratio at the receiver
J

l A RAKE is employed to do this:


It attempts to collect the time-shifted versions of the original
ll

signal by providing a separate correlation receiver for each of


the multipath signals.
A
ld
7.11 RAKE Receiver

or
The RAKE receiver is essentially a diversity receiver
designed specifically for CDMA, where the diversity is provided

W
by the fact that the multipath components are practically
uncorrelated from one another when their relative propagation
delays exceed a chip period.

TU
J N
ll

An M branch (M-finger) RAKE receiver implementation. Each correlator detects a time shifted
version of the original CDMA transmission, and each finger of the RAKE correlates to a portion
A

of the signal which is delayed by at least one chip in time from the other fingers.
ld
7.12 Interleaving

or
l Interleaving is used to obtain time diversity in a digital
communications system without adding any overhead.

W
useful technique in all second and third generation digital cellular
systems.

l It is typical for many speech coders to produce several

l
TU
"important" bits in succession.

Interleaver is employed to spread these bits out in time


so that if there is a deep fade or noise burst, the important bits
N
from a block of source data are not corrupted at the same time.
J

l Then the error control coding (called channel coding) can be


used to correct these errors.
ll

usually error control coding can deal with random errors.


A
ld
7.12 Interleaving

or
Two types of interleaver:
1) Block structure

W
l Formats the encoded data into a rectangular array of m rows
and n columns, and interleaves nm bits at a time.
l Usually, each row contains a word of source data having n bits.
An interleaver of degree m (or depth m) consists of m rows.

TU
J N
ll
A
ld
7.12 Interleaving

or
l Source bits are placed into the interleaver by sequentially

W
increasing the row number for each successive bit, and filling
the columns.

l The interleaved source data is then read out row-wise and

TU
transmitted over the channel.
This has the effect of separating the original source bits by m bit
periods.
N
l At the receiver, the de-interleaver stores the received data by
sequentially increasing the row number of each successive bit,
J

and then clocks out the data row-wise, one word (row) at a time.
ll
A
ld
7.12 Interleaving

or
Delay introduced by interleaving

W
l There is an inherent delay associated with an interleaver .
since the received message block cannot be fully decoded until
all of the nm bits arrive at the receiver and are de-interleaved.

l
than 40 ms occur.
TU
Human speech is tolerable to listen to until delays of greater

all of the wireless data interleavers have delays which do not


exceed 40ms.
N
l The interleaver word size and depth are closely related to
J

(a) the type of speech coder used


(b) the source coding rate
ll

(c) the maximum tolerable delay


A
ld
7.12 Interleaving

or
2) Convolutional structure

W
l Can be used in place of block interleavers in much the same
fashion.
l Ideally suited for use with convolutional codes.

TU
J N
ll
A
7.13 Fundamental of

ld
Channel Coding

or
l Channel coding protects digital data from errors by selectively
introducing redundancies in the transmitted data.

W
l Two types of Channel codes
1) error detection codes
2) error correction codes.

l TU
The basic purpose of Channel Coding:
Introduce redundancies in the data to improve wireless link
performance.
N
l Cost: Increases the bandwidth requirement for a fixed source
J

data rate.
This reduces the bandwidth efficiency of the link in high SNR
ll

conditions.
But provides excellent BER performance at low SNR values.
A
7.13 Fundamental of

ld
Channel Coding

or
W
Note:

Channel Coding will be detailed in another course.

TU
J N
ll
A
ld
Wireless Networks

or
W
TU
CSG 250
Spring 2005
Rajmohan Rajaraman
JN
ll
A

Wireless Networks Spring 2005


Outline of the course: Basic topics

ld
Transmission Fundamentals

or
o Analog and digital transmission

W
o Channel capacity
o Antennas, propagation modes, and fading
o Signal encoding techniques

TU
Spread spectrum technology
Coding and error control
Cellular networks
JN

Wireless LANs
o IEEE 802.11
ll

o Bluetooth
A

Wireless Networks Spring 2005


Outline: Advanced topics

ld
Mobile IP

or
TCP for wireless
Multihop ad hoc networks

W
o MAC and routing protocols
o Power control and topology control

TU
o Capacity of ad hoc networks
Sensor networks
JN
o Infrastructure, MAC, and routing protocols
o Algorithms for query processing
ll
A

Wireless Networks Spring 2005


Wireless Comes of Age

ld
 Guglielmo Marconi invented the wireless telegraph in 1896

or
o Communication by encoding alphanumeric characters in analog signal
o Sent telegraphic signals across the Atlantic Ocean

W
 Communications satellites launched in 1960s
 Advances in wireless technology
o Radio, television, mobile telephone, communication satellites
 More recently
TU
o Satellite communications, wireless networking, cellular technology, ad
hoc networks, sensor networks
JN
ll
A

Wireless Networks Spring 2005


Wireless communication systems

ld
 Target information systems: “Anytime, Anywhere, Any

or
form”
 Applications: Ubiquitous computing and information

W
access
 Market in continuous growth:
o 35-60% annual growth of PCS

TU
o Number of subscribers:
• by 2001: over 700M mobile phones
• by 2003: 1 billion wireless subscribers (source Ericsson)
JN
o 300% growth in wireless data from 1995-1997
 Large diversity of standards and products
 Confusing terminology
ll
A

Wireless Networks Spring 2005


Limitations and difficulties

ld
Wireless is convenient and less expensive

or
Limitations and political and technical difficulties

W
inhibit wireless technologies
Lack of an industry-wide standard
Device limitations
TU
o E.g., small LCD on a mobile telephone can only
displaying a few lines of text
JN
o E.g., browsers of most mobile wireless devices use
wireless markup language (WML) instead of HTML
ll
A

Wireless Networks Spring 2005


ld
or
c

W
IMT200, WLAN, ho
DAB, GSM, ad
TETRA, ...

TU
JN
Personal Travel Assistant,
PDA, laptop, GSM, cdmaOne,
WLAN, Bluetooth, ...
ll
A

Wireless Networks Spring 2005


Radio frequency spectrum

ld
Wireless technologies have gradually migrated to

or
higher frequencies

W
twisted
pair

1 Mm 10 km
coax cable

100 m
TU 1m 10 mm 100 µm
optical transmission

1 µm
JN
300 Hz 30 kHz 3 MHz 300 MHz 30 GHz 3 THz 300 THz

VLF LF MF HF VHF UHF SHF EHF infrared visible light UV


ll
A

Wireless Networks Spring 2005


Wireless & Mobility

ld
 Wireless:

or
o Limited bandwidth
o Broadcast medium: requires multiple access schemes

W
o Variable link quality (noise, interference)
o High latency, higher jitter
o Heterogeneous air interfaces
o
 Mobility:
o
TU
Security: easier snooping

User location may change with time


JN
o Speed of mobile impacts wireless bandwidth
o Need mechanism for handoff
o Security: easier spoofing
 Portability
ll

o Limited battery, storage, computing, and UI


A

Wireless Networks Spring 2005


Classification of Wireless Systems

ld
 Personal communication systems

or
o Focus on voice communication
o Limited bit-rate data transmission
o Large-scale mobility and coverage

W
o Operate over licensed frequency bands
 Wireless LANs
o Designed for high bit-rate transmission
o
o
o
IP oriented
TU
Low-scale coverage
Use unlicensed ISM frequency bands
JN
 Multihop ad hoc networks
o Have little or no infrastructure
o Low-scale coverage
o Need new routing protocols
ll

o Emerging applications
A

Wireless Networks Spring 2005


Transmission fundamentals

ld
Electromagnetic signals

or
o Time domain

W
o Frequency domain
Data rate and bandwidth
Analog and digital data transmission
TU
Channel capacity
o Nyquist theorem
JN
o Shannon capacity theorem
Transmission media
ll
A

Wireless Networks Spring 2005


ld
or
W
TU
JN
ll
A

Wireless Networks Spring 2005


Analog signaling

ld
or
W
TU
JN
ll
A

Wireless Networks Spring 2005


Digital signaling

ld
or
W
TU
JN
ll
A

Wireless Networks Spring 2005


ld
or
W
TU
JN
ll
A

Wireless Networks Spring 2005


Classification of transmission media

ld
 Transmission medium

or
o Physical path between transmitter and receiver
 Guided media

W
o Waves are guided along a solid medium
o E.g., copper twisted pair, copper coaxial cable, optical fiber
 Unguided media
TU
o Provides means of transmission but does not guide
electromagnetic signals
o Usually referred to as wireless transmission
JN
o E.g., atmosphere, outer space
ll
A

Wireless Networks Spring 2005


Unguided media

ld
Transmission and reception are achieved by

or
means of an antenna

W
Configurations for wireless transmission
o Directional
o Omnidirectional
TU
JN
ll
A

Wireless Networks Spring 2005


General frequency ranges

ld
or
 Microwave frequency range
o 1 GHz to 40 GHz
o Directional beams possible

W
o Suitable for point-to-point transmission
o Used for satellite communications
 Radio frequency range
o 30 MHz to 1 GHz
TU
o Suitable for omnidirectional applications
 Infrared frequency range
JN
o Roughly, 3x1011 to 2x1014 Hz
o Useful in local point-to-point multipoint applications within confined
areas
ll
A

Wireless Networks Spring 2005


Terrestrial microwave

ld
 Description of common microwave antenna

or
o Parabolic "dish", 3 m in diameter
o Fixed rigidly and focuses a narrow beam

W
o Achieves line-of-sight transmission to receiving antenna
o Located at substantial heights above ground level
 Applications
TU
o Long haul telecommunications service
o Short point-to-point links between buildings
JN
ll
A

Wireless Networks Spring 2005


Satellite microwave

ld
 Description of communication satellite

or
o Microwave relay station
o Used to link two or more ground-based microwave

W
transmitter/receivers
o Receives transmissions on one frequency band (uplink), amplifies
or repeats the signal, and transmits it on another frequency
(downlink)
 Applications
TU
o Television distribution
o Long-distance telephone transmission
JN
o Private business networks
ll
A

Wireless Networks Spring 2005


Broadcast radio

ld
 Description of broadcast radio antennas

or
o Omnidirectional
o Antennas not required to be dish-shaped

W
o Antennas need not be rigidly mounted to a precise alignment
 Applications
o Broadcast radio
TU
• VHF and part of the UHF band; 30 MHZ to 1GHz
• Covers FM radio and UHF and VHF television
JN
ll
A

Wireless Networks Spring 2005


Infrared

ld
Beyond the EHF spectrum

or
o 1012 to 1014 Hz

W
Transceivers must be within line of sight or
reachable via reflection
o Does not penetrate walls
TU
JN
ll
A

Wireless Networks Spring 2005

You might also like