33
EE 552/452, Spring, 2008 Wireless Communications (and Networks) Zhu Han Department of Electrical and Computer Engineering Class 10 Feb. 21 st , 2008

EE542 452 Class10 Fade

Embed Size (px)

Citation preview

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 1/33

EE 552/452, Spring, 2008Wireless Communications

(and Networks)

Zhu Han

Department of Electrical and Computer Engineering

Class 10

Feb. 21st, 2008

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 2/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

OutlineOutline

R eview: Four types of Fading ± Slow Fading

 ± Fast Fading

 ± Flat Fading

 ± Frequency Selective Fading

R ayleigh and R icean Distributions

Statistical Models

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 3/33

EE 542/452 Spring 2008

Types of Small Types of Small- -scale Fading scale Fading 

Small-scale Fading(Based on Multipath Tme Delay Spread)

Flat Fading

1. BW Signal < BW of Channel

2. Delay Spread < Symbol Period

Frequency Selective Fading

1. BW Signal > Bw of Channel

2. Delay Spread > Symbol Period

Small-scale Fading(Based on Doppler Spread)

Fast Fading

1. High Doppler Spread2. Coherence Time < Symbol Period3. Channel variations faster than baseband

signal variations

Slow Fading

1. Low Doppler Spread2. Coherence Time > Symbol Period3. Channel variations smaller than baseband

signal variations

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 4/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Fading DistributionsFading Distributions

Describes how the received signal amplitude changes with time. ±  R emember that the received signal is combination of multiple signals

arriving from different directions, phases and amplitudes.

 ± With the received signal we mean the baseband signal, namely the

envelope of the received signal (i.e. r(t)).

It is a statistical characterization of the multipath fading.

Two distributions

 ±  R ayleigh Fading

 ±  R icean Fading

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 5/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

R ay leigh DistributionsR ay leigh Distributions

Describes the received signal envelope distribution for channels, where all

the components are non-LOS:

 ± i.e. there is no line-of±sight (LOS) component.

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 6/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

R icean DistributionsR icean Distributions

Describes the received signal envelope distribution for channels where one

of the multipath components is LOS component.

 ± i.e. there is one LOS component.

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 7/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

R ay leigh Fading R ay leigh Fading 

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 8/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

R ay leigh Fading R ay leigh Fading 

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 9/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Rayleigh Fading DistributionRayleigh Fading Distribution

The R ayleigh distribution is commonly used to describe thestatistical time varying nature of the received envelope of a flatfading signal, or the envelope of an individual multipath

component.

The envelope of the sum of two quadrature Gaussian noise

signals obeys a R ayleigh distribution.

W is the rms value of the received voltage before envelopedetection, and W2 is the time-average power of the received

signal before envelope detection.

p r 

r r r 

( )exp( )

! e e g

®

¯±

°±

2

2

22

0

0 0

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 10/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Rayleigh Fading DistributionRayleigh Fading Distribution

The probability that the envelope of the received signal doesnot exceed a specified value of R  is given by the CDF:

r  peak =W and p(W)=0.6065/W

´ R R

r  edr r  p Rr  P  R P 0

2 2

2

1)()()( W 

2

)(2

1177.1

2533.12)(][

0

0

!

!!

!!!!

´

´

¡ 

rms

median

mean

dr r  pr 

dr r rpr  E r 

median

 s l if  

W W T 

W r  E r E r r p r dr  2 2 2 22

0

2

20 4292! ! !

g

´[ ] [ ] ( ) .

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 11/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

R ay leigh PDF R ay leigh PDF 

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 1 2 3 4 5W W W W W

W

mean = 1.2533W

median = 1.177W

variance = 0.4292W

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 12/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

 A typical Rayleigh fading envelope at 900MHz. A typical Rayleigh fading envelope at 900MHz.

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 13/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

R icean DistributionR icean Distribution

When there is a stationary (non-fading) LOS signal present, then theenvelope distribution is R icean.

The R icean distribution degenerates to R ayleigh when the dominantcomponent fades away.

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 14/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Ricean Fading DistributionRicean Fading Distribution

When there is a dominant stationary signal component present, the small-scale fading envelope distribution is R icean. The effect of a dominant signalarriving with many weaker multipath signals gives rise to the R iceandistribution.

The R icean distribution degenerates to a R ayleigh distribution when thedominant component fades away.

The R icean distribution is often described in terms of a parameter K which isdefined as the ratio between the deterministic signal power and the varianceof the multipath.

K is known as the R icean factor 

As Ap0, K  p -g dB, R icean distribution degenerates to R ayleighdistribution.

p r 

r r A 

A r 

r A 

( ) exp[

( )

] ( ) ,!

e e g u

®

±̄

°±W W W 2

2 2

2 0 22 0 0

0 0

K  A !2

22W 

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 15/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

C DF C DF 

Cumulative distribution for three small-scale fading measurements and their fit to R ayleigh, R icean, and log-normal distributions.

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 16/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

PDF PDF 

Probability density function of R icean distributions: K=-�dB(R ayleigh) and K=6dB. For K>>1, the R icean pdf isapproximately Gaussian about the mean.

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 17/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Rice time seriesRice time series

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 18/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

N akagami Model N akagami Model 

Nakagami Model

r: envelope amplitude

 =<r2>: time-averaged power of received signal 

m: the inverse of normalized variance of r2

 ± Get R ayleigh when m=1

m

mm

m

r m

r m

r  p;+

;

!

)(

)exp(2

)(

212

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 19/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Small Small- -scale fading mechanismscale fading mechanism

Assume signals arrive from allangles in the horizontal plane0<<360

Signal amplitudes are equal,independent of

Assume further that there is no

multipath delay: (flat fading

assumption)

Doppler shifts

nn av

 f   cosP

!

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 20/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

SmallSmall--scale fading: effect of Doppler in ascale fading: effect of Doppler in amultipath environmentmultipath environment

f m, the largest Doppler shift

2

21

8

1)( ¹¹

 º

 ¸©©ª

¨!

mm

bbEz f  

 f  k 

 f   f  S 

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 21/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

C arrier Do ppler s pectrumC arrier Do ppler s pectrum

Spectrum Empirical investigations show results that deviate

from this model Power Model Power goes to infinity at fc+/-fm

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 22/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Baseband S  pectrum Do ppler Faded Signal Baseband S  pectrum Do ppler Faded Signal 

Cause baseband spectrum has a maximum frequency of 2fm

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 23/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Simulating Do ppler/Small Simulating Do ppler/Small- -scale fading scale fading 

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 24/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Simulating Do ppler fading Simulating Do ppler fading 

Procedure in page 222

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 25/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Level C rossing R ate ( LCR)Level C rossing R ate ( LCR)

Threshold (R)

LCR is defined as the expected rate at which the Rayleigh fadingenvelope, normalized to the local rms signal level, crosses a specifiedthreshold level R in a positive going directionpositive going direction. It is given by:

secondper crossings

rms)tonormalizedvalueenvelope(specfied

where 

:

/

22

 R

r ¢  s

¢  R

 N 

r  R

e f   N 

!

!

 V

 VT V

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 26/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

 Average Fade Duration Average Fade Duration

Defined as the average period of time for which the received signal isbelow a specified level R.

For Rayleigh distributed fading signal, it is given by:

r s

 R R

 R

 f  

e

e N 

 Rr  N 

!

!

!e!

 VT  V

 V

 V

 ,2

1

11]Pr[1

2

2

Example 5.7, 5.8, 5.9

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 27/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Fading Model: Gilbert Fading Model: Gilbert- -Elliot Model Elliot Model 

Fade Period

Time t

Signal Amplitude

Threshold

Good(Non-fade)

Bad(Fade)

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 28/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Gilbert Gilbert- -Elliot Model Elliot Model 

Good(Non-fade)

Bad(Fade)

1/ANFD

1/AFD

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)

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 29/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Simulating 2 Simulating 2- -ray multi  pathray multi  path

a1 and a2 are independent R ayleigh fading

J1 and J2 are uniformly distributed over [0,2T )

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 30/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Simulating multi  path with Do ppler Simulating multi  path with Do ppler- -induced R ay leigh fading induced R ay leigh fading 

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 31/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

Saleh and Valenzuela Indoor Model Saleh and Valenzuela Indoor Model 

Measured same-floor indoor characteristics

 ± Found that, with a fixed receiver, indoor channel is very slowly time-varying

 ±  R MS delay spread: mean 25ns, max 50ns

 ± Maximal delay spread 100ns-200ns

 ± With no LOS, path loss varied over 60dB range and obeyed log distance

 power law, 3 > n > 4

Model assumes a structure and models correlated multipath components.

Multipath model ± Multipath components arrive in clusters, follow Poisson distribution.

Clusters relate to building structures.

 ± Within cluster, individual components also follow Poisson distribution.Cluster components relate to reflecting objects near the TX or R X.

 ± Amplitudes of components are independent R ayleigh variables, decayexponentially with cluster delay and with intra-cluster delay

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 32/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

SI RC IM and SM RC IM indoor/outdoor ModelsSI RC IM and SM RC IM indoor/outdoor Models

These models were developed byR 

appaport and seidel SIR 

CIM is acomputer program , that generates small scale indoor channel responsemeasurements.

The most salient feature of the model is that it produces multipath channel conditionsthat are very realistic since they are based on real world measurements and may thus be used for meaningful system design in factories and office buildings

These programs are very useful and poplar and are used in over 100institutions.

Model can measure individual multipath fading and small scale receiver spacing.

Multipath delay inside the building was found to be 40ns to 800ns.

Mean multipath delay ranged from 30-300 ns.

Arriving multipath component has a Gaussian distribution.

Average number of multipath components range from 9 to 36

8/6/2019 EE542 452 Class10 Fade

http://slidepdf.com/reader/full/ee542-452-class10-fade 33/33

EE 542/452 Spring 2008EE 552/452 Spring 2007

SI RC IM and SM RC IM indoor/outdoor ModelsSI RC IM and SM RC IM indoor/outdoor Models

SIR 

CIM Model ± Based on measurements at 1300MHz in 5 factory and other 

 buildings

 ± Model power-delay profile as a piecewise function

±±±

°

±±±

¯

®

!

ns500Tns2001360

200T-0.22

ns200ns110360

11065.0

ns110367

1

),(

K K 

1 K 

 K 

 K  K 

 K  R T T 

T T 

S T  P 

±°

±¯

!

ns500ns100)75

100T0.62exp(0.08

ns100667

55.0),( 2

 K 

 K 

 K 

 K  R

T T 

S T  P