Mobile Communications

5th Year – Telecom Engineering TRACK

1st Term – 2014/2015

MobileChannel
Mobile ChannelCharacterization
Characterization Lecture 02-v06

Communication Department
 Reminder: The Challenges…
• Propagation phenomenon:
- Severe power loss with distance (path-loss)
- Effect of large obstructing objects (shadowing)
=> slow variations of Rx power
- Multiple reflections (multipath)
=> Fading: fast variations of Rx power with time and/or frequency
• Mobility (of mobile stations or surrounding objects/reflectors):
- Time-variant channel response
- Doppler Effect
• Problem of network entry/connection continuity
• Medium sharing/Interferences
• Mobile handset limitations
- Complexity/cost
- Size/weight
- Power consumption/autonomy
• Security issues
• Health hazard & environmental considerations
HIAST – Mobile Communications Alsawah - 2013/2014 2
 In this lecture:
 Received power fluctuation phenomenon in mobile systems
 Propagation and loss mechanisms
 Concept of Link Budget
 Large-Scale variations due to Path-loss & Shadowing
 Small-Scale variations due to Multipath
 Some Path-loss models
 Log-Normal Shadowing Model
 Rayleigh & Rice Flat-Fading models
 Constellation distortion due to flat fading
 Frequency-Selective Multi-Path fading
 Definition of Coherence Time & Coherence Bandwidth

HIAST – Mobile Communications Alsawah - 2013/2014 3

 A City Walk! Time
? Position
Rx Power

Tx

HIAST – Mobile Communications Alsawah - 2013/2014 4

 Mobile Rx power fluctuation
t t
t
Rx power (dBm)

Slow component Local mean Fast variations

(Trend) (Mid-Term Average) (Instantaneous)

Large-Scale Small-Scale
Variations Variations

Time
Let’s plot power variations versus Tx-Rx distance
HIAST – Mobile Communications Alsawah - 2013/2014 5
 Rx power versus distance
Linear variation with distance
on log-log scale, why?

log d
Rx power (dBm)

Slow component
(Trend)

Small-Scale
Variations

Distance (log scale)

HIAST – Mobile Communications Alsawah - 2013/2014 6
 Area Mean
(spatial average over many locations at the same distance d)

Local Mean
(spatial average over a distance of few wavelengths)
log d
Rx power (dBm)

Contour of constant
local mean

d Distance (log scale)

HIAST – Mobile Communications Alsawah - 2013/2014 7
Area-Mean Rx Power variation with distance?

Reminder: Isotropic antenna

In free space @ distance d:
Power density (W/m2) = PT /4d 2
Power intensity (W/strad) = PT /4
Isotropic
Antenna

HIAST – Mobile Communications Alsawah - 2013/2014 8

Area-Mean Rx Power variation with distance?

Rx Pwr using 2 Isotropic antennas

Tx Ant. Rx Ant.

PT PR =?

Rx antenna
Effective Aperture

PT /4 PT /4d 2 λ2/4 PR = (λ/4d)2PT

(W/strad) (W/m 2) (m 2) (W)

HIAST – Mobile Communications Alsawah - 2013/2014 9

Area-Mean Rx Power variation with distance?

“Isotropic” Free-Space Path-Loss (FSPL)

PR = (λ/4)2 (1/d) 2 PT Valid only in “Far Field” area where d » λ

Path-loss Exponent = 2

Def. FSPL = Transmitted Power / Received Power using 2 isotropic antennas

FSPL = PT /PR = (4 d/λ)2 = (4 fc d/c)2 >1

FSPL = 10 log10 (PT /PR ) ≈ 20 log10 (d) +20 log10 (fc ) – 148 >0
(dB) (m) (Hz)
Prove that: FSPL ≈ 20 log10 (d) + 20 log10 (fc ) + 32.45
FFrriiiiss
a
(dB) (Km) (MHz)
FFoorrm muulla
Slope = 20 dB/decade
= 6 dB/octave
HIAST – Mobile Communications Alsawah - 2013/2014 10
Area-Mean Rx Power variation with distance?

FSPL (dB)

HIAST – Mobile Communications Alsawah - 2013/2014 11

Area-Mean Rx Power variation with distance?

Case of arbitrary antennas

For non-isotropic antennas: Example: Half-wave dipole max gain = 2.15 dBi

Tx Ant.
Rx Ant.
GT Tx-Rx Direction GR

PR =?
PT

PR = PT GT (c/4 fc d)2 GR
10 log10

PR = PT + GT + GR - FSPL
(dBm) (dBm) (dBi) (dBi) (dB)

HIAST – Mobile Communications Alsawah - 2013/2014 12

Area-Mean Rx Power variation with distance?

Effective-Isotropic Radiated Power (EIRP)

Def.

Example

HIAST – Mobile Communications Alsawah - 2013/2014 13

Area-Mean Rx Power variation with distance?

Other losses
In addition to FSPL, other power losses must be accounted for:
 Latms. : Absorption by atmosphere water vapor & oxygen (Negligible below 3 GHz)
 Lwthr. : Attenuation by clouds, rain, snow, fog & dust (Negligible below 10 GHz)

Antenna cables & connectors

Antenna pointing
Polarization mismatch
Antenna efficiency
Outdoor-to-Indoor penetration
Human-head absorption (mobile handset)
…

Concept of Link Budget:

HIAST – Mobile Communications Alsawah - 2013/2014 14

Area-Mean Rx Power variation with distance?

Link Budget Example

30 km
HIAST – Mobile Communications Alsawah - 2013/2014 15
 Summary & Discussion
With the FSPL model (taking into account antenna gains & other types of losses), we
succeeded to understand why Area-Mean Rx power varies linearly with distance on
log-log scales.

BUT, measurements show that:

- The slope of path-loss with distance is not 20 dB/decade!
- This slope depends on the environment (rural, urban, …)
- This slope depends of Base-Station antenna height over ground
Why?

Furthermore:
- We don’t yet understand why and how the Local-Mean Rx power varies.
- We don’t yet understand the mechanism of fast variations of instantaneous power.

We need more insight into propagation mechanisms…

HIAST – Mobile Communications Alsawah - 2013/2014 16

 Other Propagation Phenomena
Frequency band of interest [500, 3000] MHz Propagation tests began by 1932

Obstruction
Reflection Scattering Diffraction & Refraction
Absorption

 Free-Space assumption is not valid => Path-Loss model must be modified

 Effect of large obstacles like buildings => Local Mean variations (Shadowing)
 Multipath phenomenon => Small-Scale fast variations (Multipath Fading)

HIAST – Mobile Communications Alsawah - 2013/2014 17

 ?

HIAST – Mobile Communications Alsawah - 2013/2014 18

 Channel Gain Modeling
Path-loss models: Theoretical-Deterministic: Free-space, 2-Ray, …
Large-Scale
Variations Empirical with model parameter fitting: Okumura-Hata,…

(Slow) Shadowing: Theoretical-Statistical

Log-normal shadowing
Empirical-Statistical

Small-Scale
Variations Multi-path Fading: Theoretical-Statistical
Rayleigh, Rice,…
(Fast) Empirical-Statistical

HIAST – Mobile Communications Alsawah - 2013/2014 19

 2-Ray Path-Loss Model
For perfect reflection

We prove that:

Indep. of frequency!

HIAST – Mobile Communications Alsawah - 2013/2014 20

 el
od
M 2-Ray Path-Loss Model
al
tic
re
eo
Th

Pwr Gain = (Amplitude Gain)²

Gain = 1/Loss
Gain (dB) = - Loss (dB)

Path-loss Exponent = 4

Slope = 40 dB/decade

[Stuber]
HIAST – Mobile Communications Alsawah - 2013/2014 21
 el
od
M
Okumura-Hata PL Model
al
(1980)
ric
pi
Em

fc =150-1000 MHz, hb =30-200 m, hm = 1-10 m, d =1-20 Km, isotropic antennas

HIAST – Mobile Communications Alsawah - 2013/2014 22

 Other PL models?
There are many other models for:
- Extended frequency range
- Large, small cells
- Indoor coverage
- Outdoor to indoor building penetration loss
- Particular environments: factory, stadium, warehouse, hilly, tunnel,…

HIAST – Mobile Communications Alsawah - 2013/2014 23

 The Shadowing

Shadowing is the “Large-scale” Rx power variation caused by large obstacles

(compared to wavelength) such as buildings and hills that absorb, diffract,
and reflect the transmitted signal.

HIAST – Broadband Networks Alsawah - 2013/2014 24

 The Shadowing
Local Mean power gain = Random variable GSH
(spatial average of instant gain over distance of 10’s of λ)
Power Gain (dB)

Area Mean power gain GPL

(due to path-loss @ distance d)

Constructive
Shadowing

Destructive
Shadowing

d Distance (log scale)

HIAST – Mobile Communications Alsawah - 2013/2014 25
 al
tic
el re
od eo
M Th Log-Normal Shadowing Model
al/
ric
pi
Em

Measurements in diverse environments show that GSH in dB is

normally distributed with mean = GPL in dB

The shadowing standard deviation (in dB) depends on the environment (3-18 dB)

HIAST – Mobile Communications Alsawah - 2013/2014 26

 el
od
M
Path-loss + Shadowing Model
al
ric
pi
Em

d 
LdB  L0  10 log10    LSH where LSH ~  0,  SH
2
 
 d0 

Loss Loss @ a reference Path-loss Shadowing

(Local Mean) distance d0 exponent (dB)

HIAST – Mobile Communications Alsawah - 2013/2014 27

HIAST – Mobile Communications Alsawah - 2013/2014 28
HIAST – Mobile Communications Alsawah - 2013/2014 29
 Multipath: complex scenario

v2

v1

Multipath first noted by 1935

In rich multipath environments (city, indoor, etc), the received power depends on
receiver position & on time in case of mobility (of the transmitter, the receiver and/or
any other object in the neighborhood)

HIAST – Mobile Communications Alsawah - 2013/2014 30

 Multipath Fading

τ2

τ1

Multipath Fading is the “Small-scale” Rx power variation and the spectral

distortion caused by the addition of signals arriving through different paths.

HIAST – Mobile Communications Alsawah - 2013/2014 31

 Multipath Impulse Response
Large nb. of
non-resolvable
Rx Pwr(τ,t) Resolvable paths (irresolvable)
1st path paths
(scattered path)

Delay
τ
32
 Simplified Impulse Response
Path gain amplitude

τ1 τ2 Path delay τ

The path gain (in baseband) is complex random variable (magnitude+phase), why?

HIAST – Mobile Communications Alsawah - 2013/2014 33

 Multipath: Special Case 1
Non-resolvable single path with NLOS

h
I

Phasor diagram

For a large number of irresolvable paths, the resultant h is a zero-mean Complex Circular
Gaussian RV (CLT)
• The resultant magnitude is Rayleigh Distributed
• The resultant phase is uniformly distributed over [0, 2]
HIAST – Mobile Communications Alsawah - 2013/2014 34
 Complex Circular Gaussian PDF
Let: h  I  jQ  s e
j f ( I ,Q )  i , q 

with zero-mean Gaussian iid real &

imaginary parts of variance  2 each.
q
Give the PDF: i
f ( I ,Q )  i, q   ...

HIAST – Mobile Communications Alsawah - 2013/2014 35

 Rayleigh Distribution

PDF of complex gain mag.: for s > 0 and zero otherwise

(=Exponential distribution) (power gain PDF)

HIAST – Mobile Communications Alsawah - 2013/2014 36

 Rayleigh PDF
PDF

s
HIAST – Mobile Communications Alsawah - 2013/2014 37
 Rayleigh CDF
Constructive fading
CDF

(Normalized power gain)

Destructive fading
(Proba = 0.63)

20 log10(s ) (dB)
HIAST – Mobile Communications Alsawah - 2013/2014 38
 Rayleigh Fading
Constructive
fading

Normalized
RMS value
Fading gain (dB)

Destructive
fading
(63% of time)

Deep fade

Time (ms) Fading Speed: Slow/Fast?

HIAST – Mobile Communications Alsawah - 2013/2014 39
 Effect on Constellation
Rayleigh fading causes random scaling & random phase rotation of transmitted symbols

Transmitted Constellation Received Constellation

Errors

HIAST – Mobile Communications Alsawah - 2013/2014 40

 Impulse & Frequency Responses
Path power gain Frequency Response Mag. (dB)

0 (dB) 0 (dB)

τ f (Hz)
τ1

Power Delay Profile (PDP) Flat Fading

The resultant path power gain is The frequency response

exponentially distributed with magnitude is flat over all
0 dB average value frequencies but is time-variant

HIAST – Mobile Communications Alsawah - 2013/2014 41

 Multipath: Special Case 2
Non-resolvable single path with LOS

Phasor diagram

For a large number of irresolvable paths, the resultant h is a non-zero mean Complex
Circular Gaussian RV (CLT)
• The resultant magnitude is Rice Distributed
• The resultant phase is not uniformly distributed over [0, 2]

HIAST – Mobile Communications Alsawah - 2013/2014 42

 Rician PDF
  mI  2 
Received signal complex envelope: R(t )  I (t )  jQ (t ) CN   ,  I 2 
~  mQ
 



PDF of
|R(t)|

HIAST – Mobile Communications Alsawah - 2013/2014 43

 Rice versus Rayleigh PDF’s
Rice Factor

Specular (LOS) average power

K
Scattered (MP) average power

A2
K
2 2

(usually taken in dB’s)

Received signal envelope value

HIAST – Mobile Communications Alsawah - 2013/2014 44

 Fading Margins in Link Budget Power
(dBm)

Tx Power

Log-normal
shadowing

Area Mean
Pr{Local Mean Rx Pwr<Threshold 1}
Shadow margin (~ 5-10 dB)

Threshold 1

Fading margin (few dB’s)

Rx Sensitivity (dBm) Threshold 2
Rayleigh
Pr{Instant. Rx Pwr<Threshold 2} fading

Distance @ cell edge

HIAST – Mobile Communications Alsawah - 2013/2014 45
 Gaussian distribution

HIAST – Mobile Communications Alsawah - 2013/2014 46

 Multipath: Special Case 3
Resolvable faded paths with/without LOS

- Multipaths can be classified into resolvable paths

- Each resolvable path consists of a number of non-resolvable paths

HIAST – Mobile Communications Alsawah - 2013/2014 47

 Multipath Impulse & Frequency Response

Path gain |H(f)|

τ1 τ2 τ3 Path delay frequency

Concept of Frequency-Selective Fading

HIAST – Mobile Communications Alsawah - 2013/2014 48

 Freq-Selective MP: simple scenario
2-Path Model

NLO
S

LOS

Impulse Response: h(τ) = gLOS δ(τ - TLOS) + gNLOS δ(τ - TNLOS)

Frequency Response: H(f) = gLOS exp(-j2 f TLOS ) + gNLOS exp(-j2 f TNLOS)

H(f) = gLOS exp(-j2 f TLOS ) [1 + g exp(-j2 f T)]

HIAST – Mobile Communications Alsawah - 2013/2014 49

 Important parameters
Example: Indoor Power-Delay Profile

(= )

max

HIAST – Mobile Communications Alsawah - 2013/2014 50

 Important definitions
Power of k-th path

HIAST – Mobile Communications Alsawah - 2013/2014 51

 Approx.: Discrete Delay Profile
Example of urban environment @ 900 MHz
Average Rx Pwr (τ)

τ (µs)
HIAST – Mobile Communications Alsawah - 2013/2014 52
 Time-Variant Frequency Response

Fre
qu enc Coherence Bandwidth (Bc )?
y Time
Coherence Time (Tc )?

HIAST – Mobile Communications Alsawah - 2013/2014 53

 B c & Tc
Coherence Bandwidth
Bc = Frequency span over which the frequency response at a given instant of
time is almost constant (flat).
Bc = Minimum Frequency separation Δf for which the correlation coefficient
between H(f;t) and H(f+ Δf ;t) is equal to 90%.

Coherence Time
Tc = Time span over which the frequency response at a given frequency is
almost constant (time-invariant).
Tc = Minimum time separation Δt for which the correlation coefficient
between H(f;t) and H(f ; Δt) is equal to 90%.

fmax = ?
HIAST – Mobile Communications Alsawah - 2013/2014 54
 fc
Doppler Spread
Doppler Shift

fD = (v/c) fc cos
 fr = fc + fD
v
Doppler Spread

Power
v

fc freq.

Maximum Doppler Spread: fmax fmax

HIAST – Mobile Communications Alsawah - 2013/2014 55
First mobile multi-user public phone…