16582/16418 Wireless

Communication
Lecture Notes 7: Mobile Radio
Channel Modeling II
St ti ti l M
Statistical Models
d l ffor F
Fading
di
Processes
Dr. Jay Weitzen
 Contents
• Quick Review of Fading Models
• Statistical Models for Channel Fading Process
– Rayleigh
– Rician
– Nakagami
• C
Calculating
l l ti F Fade
d DDurations,
ti R
Rates,
t and
dDDepths
th
• Case Study: Characterizing the MMDS Wireless
Channel
• Combating
C b i F Fading
di
– Diversity
– Interleaving
– E
Equalization
li ti
– Rake Receiver
– OFDM
• Appendix: Introduction to MIMO for 4th generation
systems c 2007-2012 Dr. Jay Weitzen

2
Quick Review of Fading Models

• Dispersion in Time and Frequency Effect

Channel model
• In Time
Time, look at relation between multipath
spread and bit duration
– Se
Selective
ect e oor Flat
at Fading
ad g
– BW of channel vs. BW of signal
• In frequency look at Doppler Spread
relative to inverse of Bit Duration
– Fast or Slow Fading g
– Signaling rate vs. channel change rate
c 2007-2012 Dr. Jay Weitzen

3
 Types(Based
ofSmall-scale
Small-scale
Small scale
Fading g
on Multipath Tİme Delay Spread)
Fading

Flat Fading F
Frequency Selective
S l ti Fading
F di

1. BW Signal < BW of Channel 1. BW Signal > Bw of Channel

2. Delay
y Spread
p < Symbol
y Period 2. Delay Spread > Symbol Period
Small-scale Fading
(Based on Doppler Spread)
Fast Fading Slow Fading

1. Low Doppler Spread

1 High Doppler Spread
1.
2. Coherence Time > Symbol Pe
2. Coherence Time < Symbol Period
3. Channel variations smaller tha
3. Channel variations faster than baseband
signal
g variations
signal variations
c 2007-2012 Dr. Jay Weitzen

4
Impulse Response of the Fading
M lti th Model
Multipath M d l

c 2007-2012 Dr. Jay Weitzen

5
 Flat Fading

• Occurs when symbol period of the

transmitted signal is much larger than the
Delay Spread of the channel
– Bandwidth of the applied signal is narrow.

• Occurs when the amplitude of the received

signal changes with time
• For example according to Rayleigh Distribution
• May cause deep fades.
– Increase the transmit power to combat this situation.
situation
c 2007-2012 Dr. Jay Weitzen

6
 Flat Fading
s(t) r(t)
h(t,

TS

0 TS 0  0 TS+

Occurs when: BC: Coherence bandwidth

BS << BC BS: Signal bandwidth
and TS: Symbol
y period
p
TS >>  : Delay Spread

c 2007-2012 Dr. Jay Weitzen

7
 Frequency Selective Fading

• Occurs when channel multipath delay

spread is greater than the symbol period.
– Symbols face time dispersion
– Channel induces Intersymbol Interference
(ISI)
• Bandwidth of the signal s(t) is wider than
the channel impulse response
response.

c 2007-2012 Dr. Jay Weitzen

8
Frequency Selective Fading
s(t) r(t)
h(t,

 TS

0 TS 0  0 TS TS+

Causes distortion of the received baseband signal

Causes Inter-Symbol Interference (ISI)

Occurs when:
BS > BC As a rule of thumb: TS < 
and
TS < 

c 2007-2012 Dr. Jay Weitzen

9
ISI
S is
s result
esu t of
o Selective
Se ect e Fading
ad g

c 2007-2012 Dr. Jay Weitzen

10
 Fast Fading
• Due to Doppler Spread
• Rate of change of the channel characteristics
is larger than the
Rate of change of the transmitted signal
• The channel changes during a symbol period.
• The channel changes because of receiver motion.
• Coherence time of the channel is smaller than the
symbol
y period
p of the transmitter signal
g

Occurs when: BS: Bandwidth of the signal

g
BS < BD BD: Doppler Spread
and TS: Symbol Period
TS > TC TC: Coherence Bandwidth

c 2007-2012 Dr. Jay Weitzen

11
 Slow Fading
• Due to Doppler Spread
• Rate of change of the channel characteristics
is much smaller than the
Rate of change of the transmitted signal

Occurs when: BS: Bandwidth

B d idth off th
the signal
i l
BS >> BD BD: Doppler Spread
and TS: Symbol Period
TS << TC TC: Coherence Bandwidth

c 2007-2012 Dr. Jay Weitzen

12
 Different
T
Types
S
of Fading

Flat Fast
Flat Slow
Fading
Fading

Symbol Period of
Transmitting Signal

 Frequency Selective Frequency Selective

Slow Fading Fast Fading

TC
TS
Transmitted Symbol Period

With Respect To SYMBOL

S O PERIOD
O
c 2007-2012 Dr. Jay Weitzen

13
 Different Types of Fading
BS
Frequency Selective Frequency Selective
Fast Fading Slow Fading
Transmitted
B
Baseband
b d BC
Signal Bandwidth

Flat Fast Flat Slow

Fading Fading

BD
BS
Transmitted Baseband Signal Bandwidth

With Respect To BASEBAND SIGNAL BANDWIDTH

c 2007-2012 Dr. Jay Weitzen

14
Statistical Models For Small
Scale Fading
Three Major Effects: Attenuation, Long-term Fading
(Shadowing), and Short-term Fading.
Fading occurs
with distance on
order of ¼
Buildings,
B ildi wavelength
Trees, cars
obstruct signals
on a medium to
small scale:
Shadowing

Attenuation:
Signal
g
Attenuates with
Distance

c 2007-2012 Dr. Jay Weitzen

16
Fading Is the Result of Constructive and
D t
Destructive
ti Wave
W Combining
C bi i

c 2007-2012 Dr. Jay Weitzen

17
Small
S a Sca
Scale
e Fading
ad g in Space and
a d Time
e

c 2007-2012 Dr. Jay Weitzen

18
Space/Time
Space/ e Interference
te e e ce patterns
patte s

c 2007-2012 Dr. Jay Weitzen

19
 Impulse Response of a Multipath
Ch
Channell

Aican be deterministic or random complex

p Gaussian Variables
c 2007-2012 Dr. Jay Weitzen

20
Many Scatterers from same distance
results in random fading at each
distance bin

c 2007-2012 Dr. Jay Weitzen

21
Many
a y Waves
a es Combine
Co b e Due
ue to Scattering
Scatte g

c 2007-2012 Dr. Jay Weitzen

22
Real and Imaginary Parts are Gaussian
D
Due to
t Central
C t l Limit
Li it Theorem
Th

c 2007-2012 Dr. Jay Weitzen

23
 Fading Distributions

• Describes how the received signal amplitude

changes with time.
– Remember that the received signal is combination of multiple
signals arriving from different directions
directions, phases and amplitudes
amplitudes.
– With the received signal we mean the baseband signal, namely
the envelope of the received signal (i.e. r(t)).
• Its is a statistical characterization of the
multipath fading.
• Often used distributions
– Rayleigh Fading
– Ricean Fading
– Nakagami Fading
c 2007-2012 Dr. Jay Weitzen

24
 Rayleigh and Rician
Di t ib ti
Distributions
• Rayleigh Describes the received signal envelope
distribution for channels, where all the
components
p are non-LOS:
• i.e. there is no line-of–sight (LOS)
component.
• Rician Describes the received signal envelope
distribution for channels where one of the
multipath components is LOS component.
• i.e. there is one LOS component.

c 2007-2012 Dr. Jay Weitzen

25
Rayleigh Fading

c 2007-2012 Dr. Jay Weitzen

26
 Rayleigh Fading

Rayleigh distribution has the probability density function (PDF) given by:
  r2 
 
 2 2 
r
 e 
p(r)   2 (0  r  )

0 (r  0)

2 is the time average power of the received signal before envelope detection.
 is the rms value of the received voltage
g signal
g before envelope
p detection

Remember: P (average power)  Vrms

2
(see end of slides 5)

c 2007-2012 Dr. Jay Weitzen

27
 Rayleigh Fading (cont
(cont’d)
d)
The probability that the envelope of the received signal does not exceed a
specified value of R is given by the CDF:
R R2

P(R)  Pr (r  R)   p(r)dr  1 e 2 2

0


rmean  E[r ]   rp (r )dr    1.2533
0
2
rmedian
1
rmedian  1.177 found by solving 
2  p(r )dr
0

rrms  2

c 2007-2012 Dr. Jay Weitzen

28
 Rayleigh PDF
0.7


0.6
mean = 1.2533
median = 1.177
0.5
variance = 0.4292
0.4

0.3

0.2

01
0.1

0
0 1 
2 
3 
4 
5

c 2007-2012 Dr. Jay Weitzen

29
Pdf
d aand
d Cd
Cdf of
o Rayleigh
ay e g Fading
ad g

c 2007-2012 Dr. Jay Weitzen

30
The
e Envelope
e ope is
s Rayleigh
ay e g Distributed
st buted

c 2007-2012 Dr. Jay Weitzen

31
Rayleigh
ay e g Fading
ad g Margin
a g

c 2007-2012 Dr. Jay Weitzen

32
 Rayleigh
ay e g Outage Probability
obab ty
Rayleigh Fading
0
10

-1
10
bility
Outage Probab

-2
10

-3
10

-4
10

-5
10
-10 0 10 20 30 40 50

Margin (dB)

c 2007-2012 Dr. Jay Weitzen

33
Digital Communication in Rayleigh
F di
Fading iis Diffi
Difficult
lt

c 2007-2012 Dr. Jay Weitzen

34
 Ricean Distribution

• When there is a stationary (non-fading)

LOS signal present, then the envelope
distribution is Ricean
Ricean.
• The Ricean distribution degenerates to
Rayleigh when the dominant component
fades away.

c 2007-2012 Dr. Jay Weitzen

35
Rician
c a PDF

c 2007-2012 Dr. Jay Weitzen

36
Rician
c a Fading
ad g

c 2007-2012 Dr. Jay Weitzen

37
Nakagami
a aga Probability
obab ty Distribution
st but o

c 2007-2012 Dr. Jay Weitzen

38
Nakagami Shape Factor

c 2007-2012 Dr. Jay Weitzen

39
Nakagami Fading for stationary
user

c 2007-2012 Dr. Jay Weitzen

40
Level Crossing and Fade Rates

c 2007-2012 Dr. Jay Weitzen

41
 Level Crossing Rate (LCR)

Threshold (R)
( )

LCR is defined as the expected rate at which the Rayleigh fading

envelope, normalized to the local rms signal level, crosses a specified
threshold level R in a positive going direction. It is given by:
2
N R  2 f m e  
where

  R / rrms (specfied envelope value normalized to rms)

N R : crossings per second

c 2007-2012 Dr. Jay Weitzen

42
 Average Fade Duration
Defined as the average period of time for which the received signal is
below a specified level R
R.

For Rayleigh distributed fading signal, it is given by:

1
  Pr[r  R] 
NR
1
NR
1 e 2
 
2
e 1 R
 , 
fm 2 rrms

c 2007-2012 Dr. Jay Weitzen

43
ADF for Different Distributions

c 2007-2012 Dr. Jay Weitzen

44
 Fading Model –
Gilbert Elliot Model
Gilbert-Elliot
Fade Period
Signal
g
Amplitude

Threshold

Time t

Good Bad
(Non-fade) (Fade)

c 2007-2012 Dr. Jay Weitzen

45
 Gilbert-Elliot Model
1/AFD

G d
Good B d
Bad
(Non-fade) (Fade)
1/ANFD

The channel is modeled as a Two-State Markov Chain.

Each state duration is memory-less and exponentially distributed.

The rate going from Good to Bad state is: 1/AFD (AFD: Avg Fade Duration)
The rate going from Bad to Good state is: 1/ANFD (ANFD: Avg Non-Fade
Duration))

c 2007-2012 Dr. Jay Weitzen

46
 16.582
16 582 C
Case St
Study:
d
Channel Measurements for
2G MMDS and applicability to
4G LTE and WiMax
 Credits

• Based on slides from

from, Dhananjay Gore
Gore,
Stanford University
• Conducted for Sprint Broadband
Broadband, 1999
1999-
2000

c 2007-2012 Dr. Jay Weitzen

48
 Goal of Program
To characterize wireless channels
for 2G MMDS but 4G has been
deployed in this band

Channel
BTS CPE

c 2007-2012 Dr. Jay Weitzen

49
 What Is MMDS?

• MMDS (Microwave Multipoint distribution

System), is a band of frequencies at 2.5 GHz,
allocated for fixed and mobile digital
communication
i ti
– Originally viewed as a “wireless cable” system for
broadcast digital
g services
– Viewed as mostly TDD
• Business case required self installable CPE
antennas
t and
d needd tto kknow reliability
li bilit and
d
channel characteristics

c 2007-2012 Dr. Jay Weitzen

50
 Typical Scenario

Co-Channel BTS

BTS

50’-100’

Ht 8’-15’
0.1 - 4 miles

Distance to
mobile
scatterers

c 2007-2012 Dr. Jay Weitzen

51
 Scenario Dimensions
• Terrain
– Rural, Suburban, Urban, Hilly
• Antenna Configuration
– BTS, CPE antenna heights & spacing
– Polarization,, Beam-width
• Reuse Factor
– 1 and 3
• Sectorization
–3
c 2007-2012 Dr. Jay Weitzen

52
 Antenna Configurations
• BTS antenna heights
– 35’, 50’, 80’,120’ (35-120 ft)
• CPE antenna heights
– Under the eaves: 85” to 95”, (~7 ft)
– Patio of a Condominium: 130” (~10 ft)
– Rooftop: 175”
175 to 220”
220 (15-20
(15 20 ft)
• CPE antenna spacing
– 0.5
0 5 - 5 wavelengths
• Beam-width 900 at BTS and 500 at CPE

c 2007-2012 Dr. Jay Weitzen

53
 Measurement Set
Set-up
up

Ant 1 NI PCI-MIO-16E-1
Dual
Ant 1
PA Ant 1
Ant 1 Dual Rx 2 x IQ DAQ Card
custom
2 x HP 4433B custom Hi LO C++
Lo LO

Signal Pre-
Signal
Generator Signal
Signal processing
Generator Generator AD
Generator Clock
HP 4433B Matlab
10MHz / 1PPS
HP 8648C
10MHz / 1PPS Data
Divider Rubidium Analysis
Circuit Clock Rubidium Divider
Clock Circuit
PC
BTS
CCI CPE
2480 MHz
4 MHz BW

c 2007-2012 Dr. Jay Weitzen

54
Measured Channel Parameters
• Path Loss
• K-factor
• Delay Spread
• Doppler Power Spectrum
• Level Crossing Rates (LCR)
• Averageg Duration of Fade ((ADF))
• Antenna Correlation
• C/I ratios
c 2007-2012 Dr. Jay Weitzen

55
 Path-Loss
Path Loss Measurements

• Published literature (AT&T measurements)

• SU measurements only for 0.1-4 miles
• SU measurements made in multiple Bay
area locations
• SU
S measurements agree with AT&T &
measurements

SU: Stanford University

c 2007-2012 Dr. Jay Weitzen

56
 G2 MMDS Path Loss Model
Median Path Loss:
PL(dB)  A  10 log10 (d / d0 )  s  PL f  PLh
for d > d0

where
l 10 ( 4 d 0 /  )
A  20 log (free space path loss)

   a  bhb  c h  , 10 meters < hb< 80 meters

 b  (mean path loss exponent)

 is the wavelength
c 2007-2012 Dr. Jay Weitzen

57
 Path Loss Model (contd.)

• s is a lognormal shadow fading

– zero mean
– terrain dependent standard deviation
• hb is the BTS height in meters
• a, b, c are constants dependent on the
terrain category
• d o is chosen as 100m (reference distance)
• d is the distance from BTS
c 2007-2012 Dr. Jay Weitzen

58
 Correction Terms

• Frequency correction terms

PL f  5.7 log f 
 2000  f in MHz

• CPE height correction term (> 2 meters)

hCPE
PLh  10.8 log(
g( ) 1 meter < h CPE < 8 meters
2

c 2007-2012 Dr. Jay Weitzen

59
 Path Loss Scatter Plot
-60

-80

-100
SU Measurements
easu e e s
-120
h Loss [dB]

* From Erceg Model

-140
- Path

-160

-180
180

-200
-1 0 1
10 10 10
Base-Terminal Distance ((km))

c 2007-2012 Dr. Jay Weitzen

60
Mean Path Loss vs Distance
-80

Super Cell
-100

-120
Mean Path-Loss [dB]]

-140
Erceg
g
-160

-180

-200
-1 0 1
10 10 10
Base-Terminal Distance (km)

c 2007-2012 Dr. Jay Weitzen

61
 K-factor Measurements
power in fixed (mean) component
K
power in
i varying
i (scattered
( tt d) componentt
Typical Signal Envelope:
K = -10
10 dB K = 6 dB
-75 -75

-80 -80
RSL(dB)

RSL(dB)
-85 -85

-90 -90

-95 -95

-100 -100
0 20 40 60 80 100 120 140 160 180 200 0 20 40 60 80 100 120 140 160 180 200

Time ((sec)) Time ((sec))

c 2007-2012 Dr. Jay Weitzen

62
 K-factor
K factor Model

• Erceg model for K-factor

K  Fs Fh Fb K o d  u
• Fs is a seasonal factor
– 1.0; summer (leaves)
– 2.5; winter (no leaves)
• Fh is the height
g factor
– (h/3)0.46 (h is the CPE height in meters)

c 2007-2012 Dr. Jay Weitzen

63
 K-factor
K factor Model (contd.)

• Fb is the beamwidth factor

– Fb = (b/10)-0.62; (b in degrees)
d  are regression
• Ko and i coefficients
ffi i t
– Ko = 10;  = -0.5
• u is a lognormal variable
– zero mean
– std. deviation of 8.0 dB

c 2007-2012 Dr. Jay Weitzen

64
 K-factor
K factor Scatter Plot
ht = 15m, 90 deg. Rx antenna
40

30

SU Measurements
K--Factor in dB

20

From Erceg Model

10

0 hr = 3m

-10
99.9% reliability

-20
-1 0 1
10 10 10
Distance in km

c 2007-2012 Dr. Jay Weitzen

65
 K-factor
K factor and Reliability

• K-factors are highly variable

• To ensure 99.9% reliability, systems

must be designed for zero K-factor
(Rayleigh fading)

c 2007-2012 Dr. Jay Weitzen

66
 Delay Spread Model

• Spike
Spike-Plus-Exponential
Plus Exponential Model (Erceg)

P   A ( )  B  e i / o  (  i )
i 0

A, B, o and  are experimentally

determined

Trms   / 2 o
e  e   / 2 o

• Good Model for directive antennas

c 2007-2012 Dr. Jay Weitzen

67
Delay Spread Scatter Plot
(Suburban)

10
oseconds (dB)

SU Measurements
0
Spread in Micro

-5
From Erceg Model
-10
10
RMS Delay S

-15

-20

-25 -1 0 1
10 10 10
Distance in km

c 2007-2012 Dr. Jay Weitzen

68
 Doppler Power Spectrum

-122
122 -126
126

-124 -128

-126 -130
db

db
-128 -132

-130 -134

fD ~0.4Hz fD ~2Hz
-132 -136

-134 -138
-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
fD(Hz) fD(Hz)

Low Wind High Wind

Rounded Spectrum with fD~ 0.1Hz- 2Hz

c 2007-2012 Dr. Jay Weitzen

69
Level Crossing Rate (LCR)

LCR is the rate ((in sec)) at which the signal

g crosses
a certain level

Level
gnal Level

Level Crossings
g
Sig

Time c 2007-2012 Dr. Jay Weitzen

70
 LCR (measured)

LCR vs BTS antenna height for "UNDER THE EAVES" CPE

1
10

0
10
LCR (per sec)

-1
10

bts 35ft
-2 bts 50ft
10
bts 80ft

-3
10
-14 -12 -10 -8 -6 -4 -2 0 2 4 6

level (dB) w.r.t. mean power

c 2007-2012 Dr. Jay Weitzen

71
 Average Duration of Fade
(ADF)
ADF iis th
the average d
duration
ti (i(in secs)) ffor which
hi h th
the signal
i l llevell
stays below a certain threshold

t1 t2 t3 t4 t5

Level
gnal Level

ADF =  ti ))/N

Sig

N: No. of fades

0 T
Time c 2007-2012 Dr. Jay Weitzen

72
 ADF (measured)

ADF vs BTS antenna height for "UNDER THE EAVES" CPE

3
10

2
10

1
10
ADF (sec)

0
10

-1
10

-2
10 bts 35ft
bts 50ft
bts 80ft
-3
10
-14 -12 -10 -8 -6 -4 -2 0 2 4 6

level (dB) w.r.t. mean power

c 2007-2012 Dr. Jay Weitzen

73
Antenna Correlation (Spatial)

s1(t)

s2(t)

E[| s1s 2 |] - E[| s1 |]E[| s 2 |]

 s ,s 
E[(| s1 |  E[| s1 |]) 2 ]E[(| s 2 |  E[| s 2 |]) 2 ]
1 2

c 2007-2012 Dr. Jay Weitzen

74
 CPE Antenna Correlation
Coefficient vs Antenna Spacing
0.8
orrelation coefficient

0.7
BTS ht 35
35’
0.6

CPE (V. Pol)

Co

05
0.5

CPE ht 97"

0.4
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Antenna separation (wavelengths)

• 0.75 - 1 wavelength spacing adequate for under the eaves CPE

g
• 10 wavelengths sufficient for BTS antenna spacing
p g
c 2007-2012 Dr. Jay Weitzen

75
Frequency Reuse
1 1 1
3 3 3
7 2 7 2 7 2 7
9 9 9 9
8 4 8 4 8 4 8
5 5 5
1
3
6 1
3
6 1
3
6 1
3
BTS (1)
2 7 2 7 2 7 2
9 9 9
4 8 4 8 4 8 4
5 5 5 5 First Tier
1
3
6 1
3
6 1
1
3
6 1
3
6 1
3
2 7 2 7 2 7 2 7 2

8
9
4 8
9
4 8
9
4 8
9
Second Tier
5 5 5
1 6 1 6 1 6 1
3 3 3 3
2 7 2 7 2 7 2
9 9 9
4 8 4 8 4 8 7
5 5 5 9
6 1 6 1 6 1 8
3 3 3
2 2 2

Reuse Factor 3 x 9
c 2007-2012 Dr. Jay Weitzen

76
Measured C/I (Cell Edge)
Excellent Conditions
Electioneer Rd Reuse = 3 Rx Ant 1
-120
Primary C/I: 29.6447 dB
Loss [-dB]

-140
C/I = 29.6dB
Path L

-160
Primary
CCI CCI
-180
20 40 60 80 100 120 140 160
Time (s)
Electioneer Rd Reuse = 3 Rx Ant 2
-120
120
Primary C/I: 33.8216 dB
Path Loss [-dB]

-140
C/I = 33.8dB
-160
Primary
CCI CCI
-180
20 40 60 80 100 120 140 160

Reuse 3
c 2007-2012 Dr. Jay Weitzen

77
Measured C/I (Cell Edge)
Poor Conditions
Welch Rd Reuse = 3 Rx Ant 1
-120
Primary C/I: 8.3504 dB
Loss [-dB]

-140
C/I = 8.3 dB
Path L

-160
Primary
CCI CCI
-180
20 40 60 80 100 120 140
Time (s)
Welch Rd Reuse = 3 Rx Ant 2
-120
120
Primary C/I: 0.21909 dB
Path Loss [-dB]

-140
C/I = 0.21 dB
-160
Primary
CCI CCI
-180
20 40 60 80 100 120 140

Reuse 3 c 2007-2012 Dr. Jay Weitzen

78
 CDF of C/ I at the Cell Edge
(Reuse= 3 x 9)

0.8 •C/I statistics

- Randomly populate subs
he Cell Edge

0.6
- Compute path loss and
shadow loss
CDF at th

0.4
- Compute C/I
0.2 - Average over many trials
80 % coverage for cell edge

0
-10 0 10 20 30 40
C/I (dB)

c 2007-2012 Dr. Jay Weitzen

79
 Summary
y
• Over 200 hrs of measurement effort
• Measured
M d parameters
t (Path
(P th L
Loss, K
K-factor
f t
and Delay Spread) appear to conform to
AT&T results
• Consistency in new measurements of
Doppler antenna correlation
Doppler, correlation, LCR and ADF
• We feel reasonably comfortable that
measurements capture the true nature of
MMDS propagation
• More measurements planned
c 2007-2012 Dr. Jay Weitzen

80
 References
• V. Erceg et. al, “An empirically based path loss model for wireless
channels in suburban environments,” IEEE JSAC, vol. 17, no. 7, July
1999, pp. 1205-1211.
• V.
V Erceg et.al,
et al “A
A model for the multipath delay profile of fixed wireless
channels,” IEEE JSAC, vol. 17, no.3, March 1999, pp. 399-410.
• Larry J. Greenstein et.al, “A new path-gain/Delay-spread propagation
Model for digital Cellular Channels,” IEEE Trans. On Vehicular
Technology vol.
Technology, vol 46
46, no
no. 2
2, May 1997
1997.
• L.J. Greenstein, S. Ghassemzadeh, V.Erceg, and D.G. Michelson,
“Ricean K-factors in narrowband fixed wireless channels: Theory,
experiments, and statistical models,” Proceedings of WPMC’99,
Amsterdam, September 1999.
• David Parsons, “The Mobile Radio Propagation Channel,” John Wiley
and Sons, 1992.
• L. J. Greenstein and Vinko Erceg, “Gain Reductions Due to Scatter on
Wireless Paths with Directional Antennas
Antennas,” IEEE Communications
Letters, vol. 3, No. 6, June 1999.
• L.J. Greenstein et.al, “Moment-method estimation of the Ricean K-
factor,” IEEE Communications Letters, vol.3, no.6, June 1999, pp.
175-176.
c 2007-2012 Dr. Jay Weitzen

81
Diversity in Mobile Radio
Systems
Space Time Fading: Wide Beam

c 2007-2012 Dr. Jay Weitzen

83
Space time Fading, narrow
beam

c 2007-2012 Dr. Jay Weitzen

84
 Independent Paths

.4
• Space Diversity
– Multiple antenna elements separated by
decorrelation distance
distance.
• Polarization Diversity
– Two transmit or receive antennas with different
polarizations Bc
• Frequency Diversity f
– Multiple narrowband channels separated by channel
coherence bandwidth T c
• Time Diversity t
– Multiple timeslots separated by channel coherence
c 2007-2012 Dr. Jay Weitzen

time. 85
 Introduction to Diversity

• Basic Idea
– Send same bits over independent fading
paths
– Combine paths to mitigate fading
T
effects
b

t
Multiple paths unlikely to fade simultaneously
c 2007-2012 Dr. Jay Weitzen

86
 How To Maximize Diversity
• Want 2 or more signals with
approximately same average power
• Want signals
g to be uncorrelated

c 2007-2012 Dr. Jay Weitzen

87
 Combining Techniques

• Selection Combining
– Fading path with highest gain used

• Equal Gain Combining

– All p
paths cophased
p and summed with equal
q
weighting

• Maximal
M i lR Ratio
ti CCombining
bi i
– All paths cophased and summed with optimal
weighting
i hti tto maximize
i i combiner
bi output
t t SNR
c 2007-2012 Dr. Jay Weitzen

88
Maximum ratio combining
(MRC)

h1*

h1

x  y

h2

h2*

c 2007-2012 Dr. Jay Weitzen

89
Maximum
a u ratio
at o combining
co b g (cont’d)
(co t d)

c 2007-2012 Dr. Jay Weitzen

90
Selection combining (SC)

Monitor Select
SNR branch

h1

x y

h2

c 2007-2012 Dr. Jay Weitzen

91
 Switched diversity

• Switched diversity
– Switch-and-stay combining (SSC)
Channel switching
– Switch-and-examine combining (SEC) Comparator
estimator threshold

h1

h2

c 2007-2012 Dr. Jay Weitzen

92
 Calculating
Ca cu at g Probability
obab ty ofo Error
o
Introduction
• Improvements related to a reduced fading level are
commonly quantified by average error rate curves.
• The
Th average error rate t may in
i some cases be
b
difficult to evaluate analytically.

Motivation
• Quantify the severity of fading by using a measure
directly related to the fading distribution.
distribution
c 2007-2012 Dr. Jay Weitzen

93
 Diversity Performance

• Maximal
M i lR Ratio
ti CCombining
bi i (MRC)
– Optimal technique (maximizes output SNR)
– Combiner SNR is the sum of the branch SNRs
SNRs.
– Distribution of SNR hard to obtain.
– Exhibits 10-40 dB gains in Rayleigh fading.

• Selection Combining (SC)

– Combiner SNR is the maximum of the branch SNRs.
– Diminishing returns with # of antennas.
– CDF easy to obtain, pdf found by differentiating.
– Can get up to about 20 dB of gain.
c 2007-2012 Dr. Jay Weitzen

94
 Multiuser diversity Gain
System throughput for N users > than for 1 user

Spatial diversity SC 2
SEC
K

User 1

Multiuser diversity User 2

•Combiner = Base station
Antennas = Individual users
•Antennas
User K
c 2007-2012 Dr. Jay Weitzen 95
 Multi-User
u t Use Diversity
e s ty (co
(cont’d)
t d)

Introduction
• Always searching for the best user results in
a high and determinstic feedback load.
Motivation
• Utilize switched diversity algorithms reported
in the literature as multiuser access schemes
to reduce the average feedback load.
• The base station probes the users in a
sequential manner, looking not for the best
user but
b t for
f an acceptable
t bl user.
c 2007-2012 Dr. Jay Weitzen

96
Combating Rayleigh Fading:
S
Space Di
Diversity
it
D
• Fortunately, Rayleigh fades are
very short and last a small
percentage of the time
• Two antennas separated by
several wavelengths will not
generally experience fades at the
same time
• “Space Diversity” can be obtained
by using two receiving antennas
Signal received
and
d switching
it hi iinstant-by-instant
t t b i t t to t
by Antenna 1
whichever is best
• Required separation D for good
decorrelation is 10-20
Signall received
Si i d
by Antenna 2 – 12-24 ft. @ 800 MHz.
– 5-10 ft. @ 1900 MHz.
Combined
Signal
c 2007-2012 Dr. Jay Weitzen

97
Space Diversity Application
Limitations
D
• Space Diversity can be applied only
on the
th receivingi i end d off a link.
li k
• Transmitting on two antennas
would:
– fail to produce diversity
diversity, since
the two signals combine to
produce only one value of
signal
g level at a g given p point --
Signal received no diversity results.
by Antenna 1 – produce objectionable nulls in
the radiation at some angles
Signall received
Si i d • Therefore, space diversity is
by Antenna 2 applied only on the “uplink”, i.e..,
reverse path
Combined – there isn’t
isn t room for two
Signal
sufficiently separated
c 2007-2012 antennas
Dr. Jay Weitzen
on a mobile or handheld
98
 Polarization Diversity
Where Space
p Diversity
y Isn’t Convenient
• Sometimes zoning considerations or
aesthetics preclude using separate
diversity receive antennas
• Dual-polarized antenna pairs within a
single radome are becoming popular
– Environmental clutter scatters RF
energy into all possible polarizations
V+H – Differently polarized antennas receive
or signals which fade independently
\+/ – In urban environments
environments, this is almost as
good as separate space diversity
• Antenna pair within one radome can be V-
A B A B H polarized, or diagonally polarized
– Each individual array has its own
Antenna A independent feedline
Antenna B – Feedlines connected to BTS diversity
Combined inputs in the conventional way; TX
duplexing OK c 2007-2012 Dr. Jay Weitzen

99
 The Reciprocity Principle
Does it apply to Wireless?
Between two antennas, on the same
exact frequency, path loss is the
same ini both
b th directions
di ti
• But things aren’t exactly the same in
-148.21 db
cellular --
@ 870.03 MHz
– transmit and receive 45 MHz.
apart
– antenna: gain/frequency slope?
– different Rayleigh fades
up/downlink
– often, different TX & RX antennas
– RX diversity
-148.21 db • Notice also the noise/interference
@ 870.03 MHz environment may be substantially
different at the two ends
• So, reciprocity holds only in a general
-151.86
151 86 db sense for cellular
@ 835.03 MHz c 2007-2012 Dr. Jay Weitzen

100
Frequency Diversity

c 2007-2012 Dr. Jay Weitzen

101
c 2007-2012 Dr. Jay Weitzen

102
Frequency Hopping for Diversity

c 2007-2012 Dr. Jay Weitzen

103
Frequency Hopping and C/I

c 2007-2012 Dr. Jay Weitzen

104
Receive Diversity
Performance

Diversity
gain

c 2007-2012 Dr. Jay Weitzen

105
 Interleaving and De-
interleaving for Fading
Channels
 Motivation for Interleaver

• Interleaving is a form of time diversity

– Usually combined with coding to provide
protection against burst errors caused by
fading
• Viterbi Algorithm used for detection of
convolutional codes is not effective against
burst errors
errors. We add interleaver to
distribute burst error.
c 2007-2012 Dr. Jay Weitzen

107
Forward Error Correction for Fading Channels

c 2007-2012 Dr. Jay Weitzen

108
Theory of Interleaving

c 2007-2012 Dr. Jay Weitzen

109
Error Performance on Fading Channels

c 2007-2012 Dr. Jay Weitzen

110
 Block Interleaver

Original Message
Writing
00110101110000111011
0 0 1 1
Interleaver
eading

0 1 0 1
00101011001001111011
Re

1 1 0 0
Burst Error
0 0 1 1
00110101001001111011
1 0 1 1
The order of original
g Message
g is
changed by Block Interleaver.
c 2007-2012 Dr. Jay Weitzen

111
 Block
oc Deinterleaver
e e ea e

Reading Received Message

00110101001001111011
0 1 1 1
Burst Error
Writing

0 0 0 1 DeInterleaver
W

1 1 0 0 01110001110010110011

1 0 1 1
Distributed Error
0 0 1 1

c 2007-2012 Dr. Jay Weitzen

112
Example: CD Interleaving

c 2007-2012 Dr. Jay Weitzen

113
c 2007-2012 Dr. Jay Weitzen

114
Example: Satellite
C
Communications
i ti

c 2007-2012 Dr. Jay Weitzen

115
Performance with Interleaving

c 2007-2012 Dr. Jay Weitzen

116
Combating Effects of Multipath
and Fading in Wireless Systems
 What to do against ISI?

• Wideband signals:
– channel delay = many symbol periods
– heavy distortion of the received signal
signal.
• Several techniques can be applied to reduce or
get rid of ISI in wideband signal transmission
– Equalization (2nd gen)
– spread-signal
spread signal modulation (3rd gen)
– OFDM (4th gen)

c 2007-2012 Dr. Jay Weitzen

118
 Equalization

• The received signal is filtered in such a way that

ISI is eliminated or reduced.
– Ideal ISI elimination is achieved when the filter is the
inverse of the channel response.
– Clearly, the channel must be known, or accurately
estimated to perform effective equalization.
estimated, equalization
– Therefore, the equalizer needs to be trained to adapt
itself to the time-varying
y g channel in wireless systems.
y
Usually this is achieved by transmitting a training
sequence.
• E
Equalization
li ti off th
the signal
i l results
lt iin a d
decrease
c 2007-2012 Dr. Jay Weitzen

of ISI at the cost of a lower signal-to-noise ratio

(SNR)
119
 Direct sequence spread
spectrum
t
• In DS-SS modulation, the signal is multiplied with a code that
results in a signal with a much wider bandwidth than the original
information-bearing signal. In a time-dispersive multipath
channel, the spread signal replicas, which travel via different
paths, are un-correlated
un correlated if the path delays are more than one
symbol period apart from each other. After decorrelation in the
receiver, the signal replicas from different paths are combined in
a Rake receiver, thus all received energy gy is effectively
y used.
• A disadvantage of using DS-SS with high bit-rate signals is that
to achieve a sufficiently high processing gain, a very large
bandwidth is required.
q This is especially
p y the case in an indoor
environment, where the delay times between the paths are very
short, in the order of 1 ns.

c 2007-2012 Dr. Jay Weitzen

120
 OFDM
• Symbols of high bit rate signal are distributed over a
g number of subcarriers.
large
– Low symbol rate per carrier.
– Individual carrier signals see flat fading (no ISI).
• Promising techniq
technique e for ffuture
t re high bit
bit-rate
rate
applications.
• However,, it suffers from a number of problems:
p
– a very linear amplifier in the transmitter is required to prevent
signal distortion,
– accurate synchronization in the receiver is needed
needed,
– in the transmitter and receiver real-time discrete Fourier
transform (DFT) operations have to be computed.

c 2007-2012 Dr. Jay Weitzen

121
 Improving Performance of
Wireless Channels using MIMO
(the next generation of
diversity)
 MIMO is the Next generation of
Di
Diversity
i Systems
S
• Single-input, single-output (SISO) channel
No spatial diversity
• Single-input,
g p multiple-output
p p (SIMO) channel
Receive diversity
• Multiple
Multiple-input,
input, single
single-output
output (MISO) channel
Transmit diversity
• Multiple-input,
Multiple input multiple-output
multiple output (MIMO)
channel
C bi d transmit
Combined t it andd receive
i diversity
di it
c 2007-2012 Dr. Jay Weitzen

123
Introduction to the MIMO Channel

c 2007-2012 Dr. Jay Weitzen

124
Capacity of MIMO Channels

c 2007-2012 Dr. Jay Weitzen

125
 Single Input- Single Output
systems (SISO)
x(t): transmitted signal
g y(t): received signal
g(t): channel transfer function
n(t): noise (AWGN, 2)

y(t)
x(t)
y(t) = g • x(t) +
n(t)
E
Signal to noise ratio : ρ  g 2x 2

Capacity
p y : C = σlog ( )
g2(1+)
c 2007-2012 Dr. Jay Weitzen

126
Single Input- Multiple Output (SIMO
Multiple
lti l Input-
t Single
i l Output
t t (MISO
• Principle of diversity systems (transmitter/
receiver)
• +: Higher average signal to noise ratio
Robustness
• - : Process of diminishing return
Benefit reduces in the presence of
correlation
• Maximal ratio combining
• Equal gain combining
• Selection combining
c 2007-2012 Dr. Jay Weitzen

127
Transmit Diversity

c 2007-2012 Dr. Jay Weitzen

128
Transmit Diversity with Feedback

c 2007-2012 Dr. Jay Weitzen

129
TX diversity with frequency weighting

c 2007-2012 Dr. Jay Weitzen

130
TX Diversity with antenna hopping

c 2007-2012 Dr. Jay Weitzen

131
TX Diversity with channel coding

c 2007-2012 Dr. Jay Weitzen

132
Transmit diversity via delay diversity

c 2007-2012 Dr. Jay Weitzen

133
Transmit Diversity Options

c 2007-2012 Dr. Jay Weitzen

134
 MIMO Wireless Communications: Combining TX and
RX Diversity
Di it

• Transmission over Multiple Input Multiple Output (MIMO)

radio channels

• Ad
Advantages:
t I
Improved dS
Space Di
Diversity
it and
d Ch
Channell
Capacity
• Disadvantages: More complex, more radio stations and
required channel estimation
c 2007-2012 Dr. Jay Weitzen 135
 MIMO Model

T: Time index
YNT  HNM  X MT WNT
W: Noise

• Matrix Representation

– For a fixed T
c 2007-2012 Dr. Jay Weitzen 136
 Multiple Input- Multiple Output
systems (MIMO)
H11
1 1

HN1
yNx1  HNxM xMx1  nNx1
H
M N
H1M

HNM

• Average gain   E H ij
2
 , H   H
2 1

Ptotal
• Average signal to noise ratio   t t l
2
 2
c 2007-2012 Dr. Jay Weitzen

137
 Shannon capacity

  Ex H    PT 2 H 
C  log2 det  I  2 HH   log2 det  I  2
g HH  
  σ    Mσ 
  ρ H 
log2 det  I  HH 
  M 

K= rank(H): what is its range of values?

Parameters that affect the system capacity
• Signal to noise ratio 
• Distribution of eigenvalues (u) of H

c 2007-2012 Dr. Jay Weitzen

138
 Interpretation I:
The parallel channels approach
• “Proof” of capacity formula

• Singular value decomposition of H: H =

S·U·VH
• S, V: unitary matrices (VHV=I, SSH =I)
U : = diag(uk), uk singular values of H

• V/ S: input/output eigenvectors of H
• Any input along vi will be multiplied by ui
and will appear as an output along si
c 2007-2012 Dr. Jay Weitzen

139
 Vector analysis of the signals

1. The input vector x gets projected onto the

vi’ss
2. Each projection gets multiplied by a
diff
different
t gain
i ui.
3. Each appearsu along a different si.
1
<x,v1> · v1 <x,v1> u1 s1

u2
<x,v2> · v2 <x,v2> u2 s2

uK c 2007-2012 Dr. Jay Weitzen

<x,vK> · vK <x,vK> uK sK
140
 Capacity = sum of capacities

• The channel has been decomposed into

K parallel subchannels
• Total capacity = sum of the subchannel
capacities
iti
• All transmitters send the same power:
Ex=Ek
K K
C   C k   log2 1  ρ k  K
 Ek 2
C   log2 1  2 uk 
i 1 i 1
  
u E x , v  u E
i 1
2 2 2

E n, s 
k k
ρk   2
k
2
k

k
σ c 2007-2012 Dr. Jay Weitzen

141
 Interpretation
p II:
The directional approach

• Sing
Singular
lar value
al e decomposition of H:
H H=
S·U·VH
• Eigenvectors correspond to spatial directions
(beamforming)
1
(s )
1 i1

M N
(si)N

c 2007-2012 Dr. Jay Weitzen

142
Example of directional
interpretation

c 2007-2012 Dr. Jay Weitzen

143
End of Module 7