MIMO Communication Systems

Lecture 1
Wireless Channel Models

Prof. Chun-Hung Liu

Dept. of Electrical and Computer Engineering
National Chiao Tung University
Spring 2017

2017/3/2 Lecture 1: Wireless Channel Modeling 1

 Outline
Path Loss and Shadowing
(Chapter 2 in Goldsmiths Book, Chapter 2 in Tses Book)
TX-RX Signal Models
Free-Space Path Loss
Ray Tracing
Empirical Path-Loss Models

Statistical Multipath Channels (Chapter 3 in Goldsmiths

Book)
Time-Varying Channel Impulse Response
Narrowband Fading Models
Wideband Fading Models

2017/3/2 Lecture 1: Wireless Channel Modeling 2

 Path Loss and Shadowing
Introduction
Here we will characterize the variation in received signal power
over distance due to path loss and shadowing.
Path loss is caused by dissipation of the power radiated by the
transmitter as well as effects of the propagation channel
Shadowing is caused by obstacles between the transmitter and
receiver that attenuate signal power through absorption,
reflection, scattering, and diffraction
Since variations due to path loss and shadowing occur over
relatively large distances, they are sometimes referred to as large-
scale propagation effects
Variations due to multipath occur over very short distances (on
the order of the signal wavelength), so they are sometimes
referred to as small-scale propagation effects
Figure 2.1 illustrates the ratio of the received-to-transmit power
in dB versus log-distance for the combined effects of path loss,
shadowing, and multipath.
2017/3/2 Lecture 1: Wireless Channel Modeling 3
 Path Loss and Shadowing

2017/3/2 Lecture 1: Wireless Channel Modeling 4

 Transmit and Receive Signal Models
The transmitted signal can be modeled as

(2.1)

where u(t) = x(t) + jy(t) is a complex baseband signal with in-phase component
, quadrature component . . The signal u(t) is called the
complex envelope or complex low-pass equivalent signal of s(t) .

The received signal will have a similar form:

(2.2)

where the complex baseband signal v(t) will depend on the channel through which
s(t) propagates. If s(t) is transmitted through a time-invariant channel then
, where c(t) is the equivalent lowpass channel impulse response
for the channel.
2017/3/2 Lecture 1: Wireless Channel Modeling 5
 Doppler Shift
The received signal may have a Doppler shift of fD = v cos / , where is the
arrival angle of the received signal relative to the direction of motion, v is the
receiver velocity towards the transmitter in the direction of motion, and = c/fc
is the signal wavelength ( c = 3 108 m/s is the speed of light).
The geometry associated with the Doppler shift is shown in Fig. 2.2.

The Doppler frequency is then obtained

from the relationship between signal
frequency and phase:

(2.3)

2017/3/2 Lecture 1: Wireless Channel Modeling 6

 Path Loss Model
We will ignore the Doppler term in the free-space and ray tracing models since
for typical vehicle speeds (75 Km/hr) and frequencies (around 1 GHz), it is on
the order of 100 Hz
Suppose s(t) of power Pt is transmitted through a given channel, with
corresponding received signal r(t) of power Pr , where Pr is averaged over
any random variations due to shadowing.

We define the path loss of the channel as the difference in dB between the
transmitted and received signal power:
Pt (2.5)
PL dB = 10 log10 dB
Pr
In general, the dB path loss is a nonnegative number since the channel does
not contain active elements, and thus can only attenuate the signal. The dB
path gain is defined as the negative of the dB path loss: PG = PL =
10 log10 (Pr /Pt ) , which is generally a negative number.

2017/3/2 Lecture 1: Wireless Channel Modeling 7

 Free-Space Path Loss
Consider a signal transmitted through free space to a receiver located at
distance d from the transmitter.
A line-of-sight (LOS) channel: there are no obstructions between TX and
RX and the signal propagates along a straight line between the two. The
corresponding received signal is called the LOS signal or ray.
The received signal that undergoes free-space path loss

(2.6)
p
where Gl is the product of the transmit and receive antenna field
radiation patterns in the LOS direction. The phase shift e j2d/ is due to
the distance d the wave travels.
The power in the transmitted signal s(t) is , so the ratio of received to
transmitted power
p 2
The received signal power falls off P r G l
= (2.7)
inversely proportional to d2 Pt 4d
2017/3/2 Lecture 1: Wireless Channel Modeling 8
 Free-Space Path Loss
For other signal propagation models, the received signal power falls off more
quickly relative to this distance
The received signal power is also proportional to the square of the signal
wavelength . (As the carrier frequency increases, the received power decreases)
This dependence of received power on the signal wavelength is due to the
effective area of the receive antenna
Directional antennas can be designed so that receive power is an increasing
function of frequency for highly directional links
The received power can be expressed in dBm as

(2.8)

Free-space path loss is defined as the path loss of the free-space model:

(2.9)

The free-space path gain is thus

(2.10)
2017/3/2 Lecture 1: Wireless Channel Modeling 9
 Free-Space Path Loss
Example 2.1: Consider an indoor wireless LAN with MHz,
cells of radius 10 m, and nondirectional antennas. Under the free-space
path loss model, what transmit power is required at the access point
such that all terminals within the cell receive a minimum power of
10W. How does this change if the system frequency is 5 GHz?
Solution: We must find the transmit power such that the terminals at
the cell boundary receive the minimum required power. We obtain a
formula for the required transmit power as follows:
2
4d
Pt = Pr p .
Gl

2017/3/2 Lecture 1: Wireless Channel Modeling 10

 Ray Tracing
In a typical urban or indoor environment, a radio signal transmitted
from a fixed source will encounter multiple objects in the environment
that produce reflected, diffracted, or scattered copies of the transmitted
signal, as shown in Figure 2.3.
These additional copies of the
transmitted signal, called
multipath signal components,
can be attenuated in power,
delayed in time, and shifted in
phase and/or frequency from the
LOS signal path at the receiver.
The multipath and transmitted
signal are summed together at the
receiver, which often produces
distortion in the received signal
relative to the transmitted signal.

2017/3/2 Lecture 1: Wireless Channel Modeling 11

 Two-Ray Model
The two-ray model is used when a single ground reflection dominates
the multipath effect, as illustrated in Figure 2.4.

The received signal for the two-ray model is

(2.11)

where p p is the p
time delay
p of the ground reflection relative
to the LOS ray, Gl = Ga Gb , Gr = Gc Gd , R is the ground reflection
coefficient.
2017/3/2 Lecture 1: Wireless Channel Modeling 12
 Two-Ray Model
The delay spread of the two-ray model equals the delay between the
LOS ray and the reflected ray: (x + x0 l)/c
If the transmitted signal is narrowband relative to the delay ( Bu 1 )
then u(t) u(t ). With this approximation, the received power of the
two-ray model for narrowband transmission is

(2.12)

where is the phase difference between the two

received signal components.
We can show that
(2.13)

where d denotes the horizontal separation of the antennas, ht denotes the

transmitter height, and hr denotes the receiver height.

2017/3/2 Lecture 1: Wireless Channel Modeling 13

 Two-Ray Model
When d is very large compared to we can use a Taylor series to
get
(2.14)

For asymptotically large d,

Using these approximations, the received signal power is approximately

(2.17)

or, in dBm, we have

(2.18)

Thus, in the limit of asymptotically large d, the received power falls off
inversely with the fourth power of d and is independent of the wavelength
.
2017/3/2 Lecture 1: Wireless Channel Modeling 14
 Two-Ray Model
A plot of a function of distance is illustrated in Figure 2.5 for f = 900 MHz ,
and transmit power normalized
so that the plot starts at 0 dBm.

What can you learn

from this plot?

2017/3/2 Lecture 1: Wireless Channel Modeling 15

 Two-Ray Model
Example 2.2: Determine the critical distance for the two-ray model in an
urban microcell and an indoor microcell

Solution: dc = 4ht hr / = 800 meters for the urban microcell and 160
meters for the indoor system. A cell radius of 800 m in an urban
microcell system is a bit large: urban microcells today are on the order
of 100 m to maintain large capacity. However, if we used a cell size of
800 m under these system parameters, signal power would fall off as d2
inside the cell, and interference from neighboring cells would fall off
as d4 ,and thus would be greatly reduced. Similarly, 160 m is quite large
for the cell radius of an indoor system, as there would typically be many
walls the signal would have to go through for an indoor cell radius of
that size. So an indoor system would typically have a smaller cell radius,
on the order of 10-20 m.

2017/3/2 Lecture 1: Wireless Channel Modeling 16

 Ten-Ray Model
This model assumes rectilinear streets with buildings along both sides of
the street and transmitter and receiver antenna heights that are close to
street level.

Assume a narrowband model such that u(t) u(t i )for all i, then the
received power is
(2.20)

where Ri denotes the path length of the ith reflected ray and i = 2(xi l)/ .
2017/3/2 Lecture 1: Wireless Channel Modeling 17
 Empirical Path Loss Models
Most mobile communication systems operate in complex propagation
environments that cannot be accurately modeled by free-space path
loss or ray tracing.
Many path loss models have been developed to predict path loss in
typical wireless environments such as large urban macro-cells, urban
microcells, and, more recently, inside buildings
Analytical models characterize as a function of distance, so path
loss is well defined.
Empirical measurements of as a function of distance include the
effects of path loss, shadowing, and multipath.
In order to remove multipath effects, empirical measurements for
path loss typically average their received power measurements and the
corresponding path loss at a given distance over several wavelengths.
This average path loss is called the local mean attenuation (LMA) at
distance d, and generally decreases with d due to free space path loss
and signal obstructions.

2017/3/2 Lecture 1: Wireless Channel Modeling 18

 Empirical Path Loss Models
The empirical path loss PL(d) for a given environment (e.g. a city,
suburban area, or office building) is defined as the average of the LMA
measurements at distance d, averaged over all available measurements
in the given environment.
Okumura model (for signal prediction in large urban macro cells): This
model is applicable over distances of 1-100 Km and frequency ranges of
150-1500 MHz. The empirical path loss formula of Okumura at distance d
parameterized by the carrier frequency is given by

(2.28)

where is free space path loss at distance d and carrier frequency

is the median attenuation in addition to free space path loss
across all environments, is the base station antenna height gain
factor, is the mobile antenna height gain factor, and is the
gain due to the type of environment.

2017/3/2 Lecture 1: Wireless Channel Modeling 19

 Empirical Path Loss Models
Hata Model : it is an empirical formulation of the graphical path loss
data provided by Okumura and is valid over roughly the same range
of frequencies, 150-1500 MHz.
The Hata model simplifies calculation of path loss since it is a closed-
form formula and is not based on empirical curves for the different
parameters. The standard formula for empirical path loss in urban
areas under the Hata model is

(2.31)
The parameters in this model are the same as under the Okumura
model, and is a correction factor for the mobile antenna height
based on the size of the coverage area.
Piecewise Linear (Multi-Slope) Model: it is anempirical method for
approximating path loss in outdoor microcells and indoor channels.
This approximation is illustrated in Figure 2.9 for dB attenuation
versus log-distance
2017/3/2 Lecture 1: Wireless Channel Modeling 20
 Empirical Path Loss Models

In the figure, the dots represent hypothetical empirical measurements

and the piecewise linear model represents an approximation to these
measurements.

2017/3/2 Lecture 1: Wireless Channel Modeling 21

 Empirical Path Loss Models
A special case of the piecewise model is the dual-slope model as shown
in the following

(2.35)

dc is some critical distance, and are called path loss exponent.

Indoor Attenuation Factors: Indoor environments differ widely in the

materials used for walls and floors, the layout of rooms, hallways,
windows, and open areas, the location and material in obstructing
objects, and the size of each room and the number of floors.
Indoor path loss models must accurately capture the effects of
attenuation across floors due to partitions, as well as between floors.
Partition materials and dielectric properties vary widely, and thus so
do partition losses.

2017/3/2 Lecture 1: Wireless Channel Modeling 22

 Empirical Path Loss Models
Table 2.1 indicates a few examples of partition losses measured at 900-
1300 MHz

The partition loss obtained by different researchers for the same

partition type at the same frequency often varies widely, making it
difficult to make generalizations about partition loss from a specific
data set.

2017/3/2 Lecture 1: Wireless Channel Modeling 23

 Simplified Path Loss Model
The following simplified model for path loss as a function of
distance is commonly used for system design:

(2.39)

The dB attenuation is thus

(2.40)

where K is a unitless constant which depends on the antenna

characteristics and the average channel attenuation, d0 is a reference
distance for the antenna far-field, and is the path loss exponent.
The values for K, d0 and can be obtained to approximate either
an analytical or empirical model.
Due to scattering phenomena in the antenna near-field, the simplified
model is generally only valid at transmission distances d > d0 , where is
typically assumed to be 1-10 m indoors and 10-100 m outdoors.
2017/3/2 Lecture 1: Wireless Channel Modeling 24
 Simplified Path Loss Model
When the simplified model is used to approximate empirical
measurements, the value of K < 1 is sometimes set to the free space path
gain at distance assuming omnidirectional antennas:

K (dB) = 20 log10
4d0
and this assumption is supported by empirical data for free-space
path loss at a transmission distance of 100 m.

depends on the
propagation environment:
for propagation that
approximately follows a
free-space or two-ray
model is set to 2 or 4,
respectively.

2017/3/2 Lecture 1: Wireless Channel Modeling 25

 Simplified Path Loss Model
Example 2.3: Consider the set of empirical measurements of Pr /Pt given
in the table below for an indoor system at 900 MHz. Find the path loss
exponent that minimizes the MSE between the simplified model and the
empirical dB power measurements, assuming that d0 = 1 m and K is
determined from the free space path gain formula at this . Find the
received power at 150 m for the simplified path loss model with this path
loss exponent and a transmit power of 1 mW (0 dBm).

2017/3/2 Lecture 1: Wireless Channel Modeling 26

 Simplified Path Loss Model
Solution: We first set up the MMSE error equation for the dB power
measurements as

2017/3/2 Lecture 1: Wireless Channel Modeling 27

 Shadow Fading
A signal transmitted through a wireless channel will typically
experience random variation (shadow fading) due to blockage from
objects in the signal path, giving rise to random variations of the
received power at a given distance. Such variations are also caused
by changes in reflecting surfaces and scattering objects.
The most common model for characterizing this fading attenuation
is log-normal shadowing.
In the log-normal shadowing model, the ratio of transmit-to-receive
power is assumed random with a log-normal distribution
given by

(2.43)

where , is the mean of in dB and is the

standard deviation of , also in dB.

2017/3/2 Lecture 1: Wireless Channel Modeling 28

 Shadow Fading
The mean of (the linear average path gain) can be obtained by

(2.44)

The conversion from the linear mean (in dB) to the log mean (in dB) is
derived as
(2.45)

Performance in log-normal shadowing is typically parameterized by the

log mean , which is referred to as the average dB path loss and is
in units of dB.
The distribution of the dB value of is Gaussian and given by

(2.46)

2017/3/2 Lecture 1: Wireless Channel Modeling 29

 Shadow Fading
Example 2.4: In Example 2.3 we found that the exponent for the
simplified path loss model that best fits the measurements in Table 2.3
was . Assuming the simplified path loss model with this
exponent and the same K = 31.54 dB, find , the variance of log-
normal shadowing about the mean path loss based on these empirical
measurements.
Solution The sample variance relative to the simplified path loss model
with = 3.71 is

where is the path loss measurement in Table 2.3 at distance

and . Thus

2017/3/2 Lecture 1: Wireless Channel Modeling 30

 Combined Path Loss and Shadowing
Models for path loss and shadowing can be superimposed to capture
power falloff versus distance along with the random attenuation about
this path loss from shadowing.
For the combined model of path loss and shadowing, the ratio of
received to transmitted power in dB is given by:
(2.51)

where is a Gauss-distributed random variable with mean zero

and variance .
As shown in Figure 2.1, the path loss decreases linearly relative to
with a slope of 10 dB/decade, where is the path loss exponent.
Examples 2.3 and 2.4 illustrate the combined model for path loss and
log-normal shadowing based on Table 2.3, where path loss obeys the
simplified path loss model with K = 31.54 dB and path loss exponent
= 3.71 and shadowing obeys the log normal model with mean given by
the path loss model and standard deviation
2017/3/2 Lecture 1: Wireless Channel Modeling 31
 Outage Probability under Path Loss and Shadowing
In wireless systems there is typically a target minimum received
power level below which performance becomes unacceptable
(e.g. the voice quality in a cellular system is too poor to understand).
The outage probability under path loss and shadowing is
defined as the probability that the received power at a given
distance , falls below
For the combined path loss and shadowing model, the outage
probability is

(2.52)

where the Q function is defined as the probability that a Gaussian

random variable x with mean zero and variance one is bigger than z:

or

2017/3/2 Lecture 1: Wireless Channel Modeling 32

 Outage Probability under Path Loss and Shadowing
Example 2.5: Find the outage probability at 150 m for a channel based
on the combined path loss and shadowing models of Examples 2.3 and
2.4, assuming a transmit power of Pt = 10 mW and minimum power
requirement Pmin = 110.5 dBm.
Solution: We have

An outage probability of 1% is a typical target in wireless system designs.

2017/3/2 Lecture 1: Wireless Channel Modeling 33

 Cell Coverage Area
The cell coverage area in a cellular system is defined as the expected
percentage of area within a cell that has received power above a given
minimum.
Consider a base station inside a circular cell of a given radius R. All
mobiles within the cell require some minimum received SNR (Signal-
to-Noise Ratio) for acceptable performance.
The transmit power at the base station is designed for an average
received power at the cell boundary of , averaged over the shadowing
variations.
Shadowing will cause some locations within the cell to have received
power below , and others will have received power exceeding .
Figure 2.10 shows contours of constant received power based on a fixed
transmit power at the base station for path loss and average shadowing
and for path loss and random shadowing.

2017/3/2 Lecture 1: Wireless Channel Modeling 34

 Cell Coverage Area

For path loss and average

shadowing, constant power
contours form a circle around
the base station since combined
path loss and average shadowing
is the same at a uniform distance
from the base station.
For path loss and random
shadowing the contours form an
amoeba-like shape due to the
random shadowing variations
about the average.

2017/3/2 Lecture 1: Wireless Channel Modeling 35

 Cell Coverage Area
Let be the received power in dA from combined path loss and
shadowing. The total area within the cell where the minimum power
requirement is exceeded is obtained by integrating over all incremental
areas where this minimum is exceeded:

(2.55)

where 1[] denotes the indicator function.

Define Then we can have

(2.56)

The outage probability of the cell is defined as the percentage of area

within the cell that does not meet its minimum power requirement

2017/3/2 Lecture 1: Wireless Channel Modeling 36

 Cell Coverage Area
Given the log-normal distribution for the shadowing,

(2.57)
Using this outage probability in calculating C, we get

(2.58)

where (2.59)

and is the received power at the cell

boundary (distance R from the base station) due to path loss alone.
A closed-form solution for C in terms of a and b is given by

(2.60)

2017/3/2 Lecture 1: Wireless Channel Modeling 37

 Cell Coverage Area
If the target minimum received power equals the average power at the
cell boundary: then a = 0 and the coverage area simplifies
to
(2.61)

(Note that with this simplification C depends only on the ratio )

Example 2.6: Find the coverage area for a cell with the combined path loss and
shadowing models of Examples 2.3 and 2.4, a cell radius of 600 m, a base station
transmit power of , and a minimum received power
requirement of dBm and of dBm.

Solution: We first consider and check if a = 0 to determine

whether to use the full formula (2.60) or the simplified formula (2.61). We have

so we use (2.60). Evaluating a and b from (2.59) yields

2017/3/2 Lecture 1: Wireless Channel Modeling 38

 Cell Coverage Area
Substituting these into C yields

which would be a very low coverage value for an operational cellular

system (lots of unhappy customers). Now considering the less stringent
received power requirement dBm yields a = (120+114.9) /
3.65 =1.479 and the same b = 4.41. Substituting these values into C in
(2.60) yields C = .988, a much more acceptable value for coverage area.

(End of Chapter 2 in Goldsimths Book)

2017/3/2 Lecture 1: Wireless Channel Modeling 39

 Statistical Multipath Channel Models
Here, we will examine fading models for the constructive and destructive
addition of different multipath components introduced by the channel. We
model the multipath channel by a random time-varying impulse response.
If a single pulse is transmitted over a multipath channel, the received
signal will appear as a pulse train, with each pulse in the train
corresponding to the LOS component or a distinct multipath component
associated with a distinct scatter or cluster of scatters.
An important characteristic of a multipath channel is the time delay
spread it causes to the received signal.
This delay spread equals the time delay between the arrival of the first
received signal component (LOS or multipath) and the last received signal
component associated with a single transmitted pulse.
If the delay spread is small compared to the inverse of the signal
bandwidth, then there is little time spreading in the received signal.
However, when the delay spread is relatively large, there is significant
time spreading of the received signal which can lead to substantial signal
distortion.

2017/3/2 Lecture 1: Wireless Channel Modeling 40

 Time-Varying Channel Impulse Response
Recall the transmitted signal

(3.1)

The corresponding received signal is the sum of the line-of-sight (LOS)

path and all resolvable multipath components:

(3.2)

where n = 0 corresponds to the LOS path.

The unknowns in this expression are the number of resolvable multipath

components N(t), and for the LOS path and each multipath component,
its path length and corresponding delay , Doppler
phase shift and amplitude .

2017/3/2 Lecture 1: Wireless Channel Modeling 41

 Time-Varying Channel Impulse Response
We say that two multipath
components with delay and
are resolvable if their delay
difference significantly exceeds
the inverse signal bandwidth:

Multipath components that do

not satisfy this resolvability
criteria cannot be separated
out at the receiver since
and thus
these components are
nonresolvable.

2017/3/2 Lecture 1: Wireless Channel Modeling 42

 Time-Varying Channel Impulse Response
We can simplify r(t) by letting
Then the received signal can be rewritten as

(3.4)

Since is a function of path loss and shadowing while depends on

delay and Doppler, we typically assume that these two random processes are
independent.
The received signal r(t) is obtained by convolving the baseband input
signal u(t) with the equivalent lowpass time-varying channel impulse
response c(, t) of the channel and then upconverting to the carrier
frequency
(3.5)

2017/3/2 Lecture 1: Wireless Channel Modeling 43

 Time-Varying Channel Impulse Response
We know that c(, t) must be given by

(3.6)

where c(, t) represents the equivalent lowpass response of the channel at

time t to an impulse at time t- . So it follows that

2017/3/2 Lecture 1: Wireless Channel Modeling 44

 Time-Varying Channel Impulse Response
Some channel models assume a continuum of multipath delays, in
which case c(, t) becomes

(3.7)

Figure 3.2 shows each multipath component corresponds to a single

reflector.

2017/3/2 Lecture 1: Wireless Channel Modeling 45

 Time-Varying Channel Impulse Response

2017/3/2 Lecture 1: Wireless Channel Modeling 46

 Time-Varying Channel Impulse Response
Example 3.1: Consider a wireless LAN operating in a factory near a conveyor
belt. The transmitter and receiver have a LOS path between them with gain ,
phase and delay . Every seconds a metal item comes down the conveyor
belt, creating an additional reflected signal path in addition to the LOS path with
gain , phase and delay . Find the time-varying impulse response c(,t) of
this channel.
Solution: For the channel impulse response
corresponds to just the LOS path. For the channel impulse
response has both the LOS and reflected paths. Thus, c(,t) is given by

Note that for typical carrier frequencies, the nth multipath component
will have . For example, with GHz and ns
(a typical value for an indoor system), .

2017/3/2 Lecture 1: Wireless Channel Modeling 47

 Time-Varying Channel Impulse Response
If fc n (t) 1 then a small change in the path delay can lead to a
very large phase change in the nth multipath component with phase

Rapid phase changes in each multipath component gives rise to constructive

and destructive addition of the multipath components, which causes rapid
variation in the received signal strength. This phenomenon is called fading.
The impact of multipath on the received signal depends on whether the
spread of time delays associated with the LOS and different multipath
components is large or small relative to the inverse signal bandwidth.
If the channel delay spread is small then the LOS and all multipath
components are typically nonresolvable, leading to the narrowband
fading model.
If the delay spread is large then the LOS and all multipath components
are typically resolvable into some number of discrete components, leading
to the wideband fading model.
2017/3/2 Lecture 1: Wireless Channel Modeling 48
 Narrowband Fading Models
Suppose the delay spread of a channel is small relative to the inverse
signal bandwidth B of the transmitted signal, i.e. .
Under most delay spread characterizations implies that the
delay associated with the ith multipath component so
. Thus,
In order to characterize the random scale factor caused by the multipath
we choose s(t) to be an unmodulated carrier with random phase offset

which is narrowband for any .

With this assumption the received signal becomes

(3.13)

2017/3/2 Lecture 1: Wireless Channel Modeling 49

 Narrowband Fading Models
where the in-phase and quadrature components are given by
and

where

2017/3/2 Lecture 1: Wireless Channel Modeling 50

 Narrowband Fading Models

Envelope and Power Distributions: For any two Gaussian random

variables X and Y , both with mean zero and equal variance , it can
be shown that is Rayleigh-distributed and is
exponentially distributed.
2017/3/2 Lecture 1: Wireless Channel Modeling 51
 Envelope and Power Distributions
For uniformly distributed, rI and rQ are both zero-mean
Gaussian random variables. If we assume a variance of for both
in-phase and quadrature components then the signal envelope
is Rayleigh-distributed with distribution

(3.32)

where is the average received signal power of the signal,

i.e. the received power based on path loss and shadowing alone.
We can obtain the received signal power is exponentially distributed with
mean :
(3.33)

Example 3.2: Consider a channel with Rayleigh fading and average

received power dBm. Find the probability that the received
power is below 10 dBm.
2017/3/2 Lecture 1: Wireless Channel Modeling 52
 Envelope and Power Distributions
Solution: We have . We want to find the probability
that . Thus

If the channel has a fixed LOS component, and are not zero-mean.
In this case, the received signal equals the superposition of a complex
Gaussian component and a LOS component, and the signal enveloep can
be shown to have a Rician distribution, given by

(3.34)

where is the average power in the non-LOS multipath

components and is the power in the LOS component. The
function is the modified Bessel function of 0th order.

2017/3/2 Lecture 1: Wireless Channel Modeling 53

 Envelope and Power Distributions
The average received power in the Rician fading is given by

(3.35)

The Rician distribution is often described in terms of a fading

parameter K, defined by

K is the ratio of the power in the LOS component to the power in the
other (non-LOS) multipath components. For K = 0 we have Rayleigh
fading, and for K = we have no fading, i.e. a channel with no multipath
and only a LOS component.
The fading parameter K is therefore a measure of the severity of the
fading: a small K implies severe fading, a large K implies more mild
fading.

2017/3/2 Lecture 1: Wireless Channel Modeling 54

 Envelope and Power Distributions
Nakagami fading distribution was developed based on a variety of
empirical measurements. This distribution is given by

(3.38)

where is the average received power and () is the Gamma function.

For m = 1, the Nakagami distribution reduces to Rayleigh fading. For
the , it is approximately Rician fading with K. For m
= , there is no fading: is a constant.
The power distribution for Nakagami fading, obtained by a change of
variables, is given by

(3.39)

2017/3/2 Lecture 1: Wireless Channel Modeling 55

 Wideband Fading Models
When the signal is not narrowband we get another form of distortion due
to the multipath delay spread.

In this case, a short transmitted pulse of duration T will result in a

received signal that is of duration T + Tm , where Tm is the multipath
delay spread. Thus, the duration of the received signal may be
significantly increased. This is illustrated in Figure 3.11.

2017/3/2 Lecture 1: Wireless Channel Modeling 56

 Wideband Fading Models
If the multipath delay spread Tm T then the multipath components
are received roughly on top of one another, as shown on the upper
right of the figure.
The resulting constructive and destructive interference causes
narrowband fading of the pulse, but there is little time-spreading of the
pulse and therefore little interference with a subsequently transmitted
pulse.
If the multipath delay spread Tm T , then each of the different
multipath components can be resolved, as shown in the lower right of
the figure. However, these multipath components interfere with
subsequently transmitted pulses. This effect is called intersymbol
interference (ISI).
ISI mitigation is not necessary if Tm T , but this can place significant
constraints on data rate.

2017/3/2 Lecture 1: Wireless Channel Modeling 57

 Wideband Fading Models
The difference between wideband and narrowband fading models is that
as the transmit signal bandwidth B increases so that , the
approximation is no longer valid. Thus, the received
signal is a sum of copies of the original signal, where each copy is
delayed in time by and shifted in phase by .
The received signal is a sum of copies of the original signal, where each
copy is delayed in time by and shifted in phase by .
Although the received signal in narrowband fading no longer applies when
the signal bandwidth is large relative to the inverse of the multipath delay
spread, if the number of multipath components is large and the phase of
each component is uniformly distributed then the received signal will still be
a zero-mean complex Gaussian process with a Rayleigh-distributed envelope.

However, wideband fading differs from narrowband fading in terms of the

resolution of the different multipath components.

2017/3/2 Lecture 1: Wireless Channel Modeling 58

 Wideband Fading Models
The starting point for characterizing wideband channels is the equivalent
lowpass time-varying channel impulse response c(, t) .
We can take the Fourier transform of c(, t) with respect to t as

(3.49)

We call the deterministic scattering function of the lowpass

equivalent channel impulse response c(, t) .
The statistical characterization of c(, t) is thus determined by its
autocorrelation function, defined as
(3.50)
Most channels in practice are wide-sense stationary (WSS), such that
the joint statistics of a channel measured at two different times t and
t + t depends only on the time difference t .

2017/3/2 Lecture 1: Wireless Channel Modeling 59

 Wideband Fading Models
We will assume that our channel model is WSS, in which case the
autocorrelation becomes independent of t:
(3.51)
Uncorrelated scattering (US): If a channel has a US, then the channel
response associated with a given multipath component of delay is
uncorrelated with the response associated with a multipath component at
a different delay .
Incorporating the US property into yields

(3.52)

where gives the average output power associated with the channel
as a function of the multipath delay and the difference t in
observation time.

2017/3/2 Lecture 1: Wireless Channel Modeling 60

 Wideband Fading Models
The scattering function for random channels is defined as the Fourier
transform of Ac ( ; t) with respect to the t parameter:

(3.53)

2017/3/2 Lecture 1: Wireless Channel Modeling 61

 Coherent Bandwidth
The time-varying multipath channel in the frequency domain can be
characterized by taking the Fourier transform of c( ; t) with respect to .
Specifically, define the random process

(3.57)

Since c( ; t) is WSS, its integral C(f ; t) is as well. Thus, the

autocorrelation of C(f ; t) is given by
(3.58)

(3.59)
2017/3/2 Lecture 1: Wireless Channel Modeling 62
 Coherent Bandwidth
where f = f2 f1 and the third equality follows from the WSS and US
properties of c( ; t) .
The autocorrelation of C(f ; t) in frequency depends only on the frequency
difference f .
(3.60)

The frequency Bc where Ac ( f ) 0 for all f > Bc is called the

coherence bandwidth of the channel.

By the Fourier transform relationship between , if

the then for . Thus, the
minimum frequency separation for which the channel response is
roughly independent is , where T is typically taken to be the rms
delay spread .

2017/3/2 Lecture 1: Wireless Channel Modeling 63

 Coherent Bandwidth
In general, if we are transmitting a narrowband signal with bandwidth
B Bc , then fading across the entire signal bandwidth is highly
correlated, i.e. the fading is roughly equal across the entire signal
bandwidth. This is usually referred to as flat fading.
If the signal bandwidth B Bc , then the channel amplitude values at
frequencies separated by more than the coherence bandwidth are
roughly independent. Thus, the channel amplitude varies widely across
the signal bandwidth. In this case, the channel is called frequency-
selective.
Note that in linear modulation the signal bandwidth B is inversely
proportional to the symbol time Ts , so flat fading corresponds
to Ts 1/B 1/Bc = Tm , i.e. the case where the channel experiences
negligible ISI.
Frequency-selective fading corresponds to Ts 1/B 1/Bc = Tm , i.e.
the case where the linearly modulated signal experiences significant ISI.

2017/3/2 Lecture 1: Wireless Channel Modeling 64

 Coherent Bandwidth
The power delay profile Ac ( ) and its Fourier transform AC ( f ) in
Figure 3.13.

2017/3/2 Lecture 1: Wireless Channel Modeling 65

 Coherent Bandwidth
Example 3.6: In indoor channels ns whereas in outdoor
microcells sec. Find the maximum symbol rate Rs = 1/Ts
for these environments such that a linearly-modulated signal
transmitted through these environments experiences negligible ISI.

Solution: We assume that negligible ISI requires Ts Tm , i.e. Ts 10 Tm .

This translates to a symbol rate Rs = 1/Ts .1/ Tm .
For ns this yields Mbps and for Tm 30 sec this yields
Rs 3.33 Kbps. Note that indoor systems currently support up to 50 Mbps
and outdoor systems up to 200 Kbps. To maintain these data rates for a
linearly-modulated signal without severe performance degradation due to
ISI, some form of ISI mitigation is needed. Moreover, ISI is less severe in
indoor systems than in outdoor systems due to their lower delay spread
values, which is why indoor systems tend to have higher data rates than
outdoor systems.

2017/3/2 Lecture 1: Wireless Channel Modeling 66

Doppler Power Spectrum and Channel Coherence Time
A Doppler shift in the received signal is caused by the time variations of
the channel which arise from transmitter or receiver motion.
It can be characterized by taking the Fourier transform of AC ( f ; t)
relative to t:
(3.61)

In order to characterize Doppler at a single frequency, we set f to zero

and define SC () , SC (0; ) which is given by

(3.62)

where AC ( t) , AC ( f = 0; t). Note that AC ( t) is an autocorrelation

function defining how the channel impulse response decorrelates over time.

AC ( t = T ) = 0 indicates that observations of the channel impulse response

at times separated by T are uncorrelated and therefore independent since
the channel is a Gaussian random process.
2017/3/2 Lecture 1: Wireless Channel Modeling 67
Doppler Power Spectrum and Channel Coherence Time
The channel coherence time Tc to be the range of values over which
AC ( t) is approximately nonzero. Thus, the time-varying channel
decorrelates after approximately Tc seconds.
The function is called the Doppler power spectrum of the
channel: as the Fourier transform of an autocorrelation it gives the PSD
of the received signal as a function of Doppler .
The maximum value for which is greater than zero is called
the Doppler spread of the channel, and is denoted by . ( )
If the transmitter and reflectors are all stationary and the receiver
is moving with velocity v, then
Recall that in the narrowband fading model samples became independent
at time , so in general where k depends on the
shape of .

2017/3/2 Lecture 1: Wireless Channel Modeling 68

Doppler Power Spectrum and Channel Coherence Time
Figure 3.14 illustrates the Doppler power spectrum SC () and its inverse
Fourier transform AC ( t) .

Example 3.7: For a channel with Doppler spread Bd = 80 Hz, what time
separation is required in samples of the received signal such that the
samples are approximately independent.
Solution: The coherence time of the channel is Tc 1/Bd = 1/80 , so samples
spaced 12.5 ms apart are approximately uncorrelated and thus these
samples are approximately independent.
2017/3/2 Lecture 1: Wireless Channel Modeling 69