Short Course on MIMO Systems

Diversity in Communications

Raviraj S. Adve
University of Toronto

Dept. of Elec. and Comp. Eng.

Toronto, Ontario, Canada M5S 3G4

[email protected]

Waveform Diversity and Design Conference, Slide #1

 Multiple Input Multiple Output Systems

MIMO Systems: the use of an antenna array at the receiver

Introduction and Overview (Multiple Output) and/or the transmitter (Multiple Input) in
Basic Digital Communications wireless communications
Basic Wireless Communications

Receive Diversity

Transmit Diversity

MIMO Information Theory

Outline of this Course:
Course Summary

END Basic digital and wireless communications

Diversity on Receive
Diversity on Transmit
Multiplexing and data rate

Detailed notes available at

http://www.comm.utoronto.ca/ rsadve/teaching.html

Waveform Diversity and Design Conference, Slide #2

 I like this research area because...

Introduction and Overview

Basic Digital Communications

MIMO Systems bring ....and it is useful!
Basic Wireless Communications
together....
Receive Diversity

Transmit Diversity
Antenna array theory
MIMO Information Theory

Course Summary
Probability theory
END
Linear algebra
Optimization
Digital communications

(borrowed from The Economist,

April 28th - May 4th 2007 )

Waveform Diversity and Design Conference, Slide #3

 A Digital Communication System

Transmitter

Introduction and Overview

Analog Sample Bits Channel Symbols Pulse x(t)
Basic Digital Communications Transmitter
Source and D/A Encoder Modulator
BER
Info Theory

Basic Wireless Communications

Receive Diversity
Channel
Transmit Diversity

MIMO Information Theory

Receiver
Course Summary Analog
Output Bits Source Symbols Demodulator y(t)
END D/A Detector
Decoder and sample

x(t) = s(t) + n(t)

The noise term, n(t), is usually modelled as additive, white,

Gaussian noise (AWGN)

Waveform Diversity and Design Conference, Slide #4

 Performance Measure : BER
Performance of the system is generally measured via the bit
error rate (BER)
Introduction and Overview
BER is a function of signal-to-noise ratio (SNR)
Basic Digital Communications

E{|s(t)|2 }
BER
Info Theory
SNR =
Basic Wireless Communications E{|n(t)|2 }
Receive Diversity

Transmit Diversity
where E{} is the expectation operator.
MIMO Information Theory Binary Phase Shift Keying (BPSK): bit = 0 symbol = 1, bit
Course Summary = 1 symbol = -1
END
For an AWGN channel and BPSK
 
BER = Q 2SNR
Z
1 2
Q(x) = et /2 dt
2 x
x2 /2
Q(x) < (1/2)e , i.e., for an AWGN channel BER falls
off exponentially with SNR.

Waveform Diversity and Design Conference, Slide #5

 Basic Information Theory
Channels are characterized by channel capacity C
Introduction and Overview Shannon says: Given a channel with capacity C, one can
Basic Digital Communications
BER
find a coding scheme to transmit at a data rate R < C
Info Theory without error. Furthermore, one cannot transmit without error
Basic Wireless Communications at a data rate R > C.
Receive Diversity
C acts as the effective speed limit on the channel
Transmit Diversity

MIMO Information Theory

R is generally measured in bits per channel use.
Course Summary
For an AWGN channel (with complex inputs and outputs)
END

C = log2 (1 + SNR) (bits)

Note that C is a non-linear function of SNR

At low SNR, C SNR
At high SNR C log2 (SNR)

Waveform Diversity and Design Conference, Slide #6

 A Wireless Communication System

Transmitter

Introduction and Overview

Data Source Bits Symbols
Channel Pulse RF
Basic Digital Communications Source Encoder Encoder Modulator Modulator
Transmitter
Basic Wireless Communications
Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading
Slow/Flat fading
Channel
Summary Impulse
SISO Data Model Response
Channel Fluctuation
Error Rate
Outage Probability

Receiver
Receive Diversity

Transmit Diversity
Format Source Bits Channel Symbols Demodulator
Data Detector
MIMO Information Theory Decoder Decoder and sample
Receiver
Course Summary

END

A wireless communication system is fundamentally limited by

the random channel, i.e., fading.

Waveform Diversity and Design Conference, Slide #7

 Fading Channels
Due to the unknown location of the mobile station and the
unknown medium between the transmitter and receiver, the
Introduction and Overview
wireless channel is best characterized as random.
Basic Digital Communications

Basic Wireless Communications

Fading Channels Fading has three components:
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading Overall fading = (Distance Attenuation) (Large Scale
Slow/Flat fading
Summary Fading) (Small Scale Fading)
SISO Data Model
Channel Fluctuation
Error Rate
Distance attenuation: fall off in power with distance
Outage Probability

Receive Diversity
In line-of-sight conditions, received power 1/d2 (where
d is the distance between transmitter and receiver)
Transmit Diversity

MIMO Information Theory

In non line-of-sight conditions, received power 1/dn ,
Course Summary where n is the distance attenuation parameter
END 1.5 < n < 4.5.
In some scenarios n < 2 due to tunnelling
n large in urban areas, 4.5

Waveform Diversity and Design Conference, Slide #8

 Fading Channels (cont...)
Large Scale Fading: Occurs due to the attenuation each time
a signal passes through an object
Introduction and Overview

Basic Digital Communications Large Scale Fading = 10(x1 +x2 +...) = 10x
Basic Wireless Communications
Fading Channels
Small Scale Fading
where xi is the attenuation due to object # i.
Rayleigh/Rician Fading
Flat/Selective Fading By the Cental Limit Theorem, x = (x1 + x2 + . . .) is a
Slow/Flat fading
Summary Gaussian random variable
SISO Data Model
Channel Fluctuation
Error Rate
large scale fading can be modelled as log-normal, i.e., the
Outage Probability log of the fading term is distributed normal:
Receive Diversity

Transmit Diversity
Large Scale Fading = 10x ;
2
MIMO Information Theory x N (0, h1 )
Course Summary

END
Large scale fading varies slowly, remaining approximately
constant over hundreds of wavelengths
There is not much one can do about large scale fading or
distance attenuation, just power control

Waveform Diversity and Design Conference, Slide #9

 Small Scale Fading
Occurs due to multipath propagation
Each signal arrives over many many paths
Introduction and Overview

Basic Digital Communications

Basic Wireless Communications

Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading
Slow/Flat fading
Summary
SISO Data Model
Channel Fluctuation
Error Rate Transmitter Receiver
Outage Probability

Receive Diversity

Transmit Diversity

MIMO Information Theory

Course Summary

END

Each path has slightly different length slightly different

time of propagation
with different phase that is effectively random
Waveform Diversity and Design Conference, Slide #10
 Small Scale Fading (cont...)
Total received signal is the sum over all paths

Introduction and Overview Received Signal = Transmitted Signal Distance Attenuation

Basic Digital Communications
Large Scale Fading (h1 + h2 . . .)
Basic Wireless Communications
Fading Channels hi = i eji ; i is random
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading If all i are approximately the same (Rayleigh fading),
Slow/Flat fading
Summary
SISO Data Model
Channel Fluctuation
Small Scale Fading = h = (h1 + h2 + . . .)
Error Rate
Outage Probability h CN (0, h2 )
Receive Diversity

Transmit Diversity
If one of the components is dominant (Rician fading)
MIMO Information Theory
h CN (, h2 )
Course Summary

END
CN (, h2 ) represents the complex Gaussian distribution with
mean and variance h2 .
We will focus on Rayleigh fading, i.e., h CN (0, h2 )
Note: Rayleigh fading is just one of many fading models
Waveform Diversity and Design Conference, Slide #11
 Small Scale Fading (cont...)
Frequency flat versus frequency selective fading:
Symbol period: Ts ; Channel time spread: Tc
Introduction and Overview

Basic Digital Communications h(t) h(t)

Modelled As
Basic Wireless Communications
Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading
Slow/Flat fading
Ts 2Ts T Ts 2Ts T
Summary c c
SISO Data Model
Channel Fluctuation
Error Rate
Outage Probability
Channel modelled as a train of impulses
Receive Diversity

Transmit Diversity
If Tc < Ts , h(t) = (t) H(j) = constant (Frequency
MIMO Information Theory
flat fading)
P
Course Summary
If Tc > Ts , h(t) = (t Ts ) H(j) 6= constant
END (Frequency selective fading)
Note: CN (0, h2 )

Waveform Diversity and Design Conference, Slide #12

 Small Scale Fading (cont...)
If the mobile is moving with radial velocity v, the channel
changes as a function of time
Introduction and Overview
Rate of change = Doppler frequency (fd = v/)
Basic Digital Communications

Basic Wireless Communications

Symbol rate = fs = 1/Ts
Fading Channels
Small Scale Fading If fd fs , channel is effectively constant over several
Rayleigh/Rician Fading
Flat/Selective Fading symbols (slow fading)
Slow/Flat fading
Summary
SISO Data Model
If fd > fs , channel changes within a symbol period (fast
Channel Fluctuation
Error Rate
fading)
Outage Probability
We shall focus on slow, flat, fading
Receive Diversity

Transmit Diversity

MIMO Information Theory

Course Summary

END

Waveform Diversity and Design Conference, Slide #13

 In summary...
Fading has three components
Introduction and Overview
Distance attenuation 1/dn
Basic Digital Communications Large scale fading; modelled as log-normal; constant over
Basic Wireless Communications hundreds of
Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Small scale fading; modelled as Rayleigh, i.e., complex
Flat/Selective Fading normal; fluctuates within fraction of
Slow/Flat fading
Summary
SISO Data Model Assume power control for distance attenuation and large
Channel Fluctuation
Error Rate scale fading.
Outage Probability
This is all that can be done!
Receive Diversity

Transmit Diversity

MIMO Information Theory

Course Summary

END

Waveform Diversity and Design Conference, Slide #14

 Data Model
With flat, slow fading:
Introduction and Overview
x = hs + n
Basic Digital Communications

Basic Wireless Communications

Fading Channels
Small Scale Fading
x: received signal, h: channel, s: transmitted complex
Rayleigh/Rician Fading
Flat/Selective Fading
symbol, n: noise
Slow/Flat fading
Summary h CN (0, h2 )
SISO Data Model
2 2
Instantaneous channel power: |h|2 (1/h2 )e|h| /h
Channel Fluctuation
Error Rate
Outage Probability
(exponential)
Receive Diversity

Transmit Diversity
channel is often in bad shape
MIMO Information Theory Note: h2 = E{|h|2 } = average power in channel
Course Summary
Set h2 = 1 for convenience (channel does not
END
introduce power)

Waveform Diversity and Design Conference, Slide #15

 Impact of Fading
E{|s|2 } 2 2
Average SNR = E{|h| } 2
= h =
Introduction and Overview

2
2 E{|s| } 2 1 /
Basic Digital Communications
= Instantaneous SNR = |h| 2
= |h| ; e
Basic Wireless Communications
Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Note that the average SNR has not changed
Flat/Selective Fading
Slow/Flat fading
Summary
The fluctuation in power due to the fading seriously impacts
SISO Data Model on the performance of a wireless system
Channel Fluctuation
10
Error Rate
Outage Probability
5
Receive Diversity

0
Instantaneous Channel power (dB)

Transmit Diversity

MIMO Information Theory 5

Course Summary
10

END
15

20

25
0 10 20 30 40 50 60 70 80 90 100
Sample #

Waveform Diversity and Design Conference, Slide #16

 Bit Error Rate

Without fading, BER = Q( 2) exp()
Exponential drop off with SNR
Introduction and Overview

Basic Digital Communications

fading, instantaneous SNR = , i.e., instantaneous BER
With
Basic Wireless Communications
= Q( 2)
Fading Channels
Small Scale Fading p
Rayleigh/Rician Fading
Flat/Selective Fading
Average BER = E {Q( 2}
Slow/Flat fading
Summary

1
SISO Data Model
Z
Q( 2) e/ d
p
Channel Fluctuation
Error Rate
=
Outage Probability 0
Receive Diversity r !
1
Transmit Diversity
= 1
MIMO Information Theory 2 1+
Course Summary

END
At high SNR ( ),
1
BER

Waveform Diversity and Design Conference, Slide #17

 Bit Error Rate: Example
Comparing error rates with and without fading
0
10
Without fading
Introduction and Overview With Rayleigh fading

Basic Digital Communications

1
10
Basic Wireless Communications
Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading 2
10
Slow/Flat fading
Summary
BER

SISO Data Model

Channel Fluctuation
Error Rate 3
10
Outage Probability

Receive Diversity

Transmit Diversity 4
10

MIMO Information Theory

Course Summary

END
0 5 10 15
SNR (dB)

Note: In the case with fading, the BER v/s SNR plot (in log-log
format appears as a straight line)

Waveform Diversity and Design Conference, Slide #18

 Outage Probability
2


If channel is known, capacity: C = log2 1 + |h|

Introduction and Overview

If channel is unknown, true capacity = 0!!
Basic Digital Communications One cannot guarantee any data rate
Basic Wireless Communications
Fading Channels
Therefore, define outage probability for a rate R
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading
Pout = P (C < R)
Slow/Flat fading
R
 
Summary 2 1
1 + |h|2 < R = P |h|2 <



SISO Data Model P log2
Channel Fluctuation
Error Rate
Outage Probability
 R 
2 1
Receive Diversity
= 1 exp
Transmit Diversity
MIMO Information Theory
IMPORTANT: as average SNR gets large ( ),
Course Summary
 R
2R 1

END 2 1
exp 1

1
Pout

Waveform Diversity and Design Conference, Slide #19
 Outage Probability: Alternate Definition
Note: Choosing a target rate R is equivalent to choosing a
SNR threshold, s
Introduction and Overview

Basic Digital Communications

Alternate definition of outage
Basic Wireless Communications
Fading Channels
Pout = P [ < s ]
Small Scale Fading
Rayleigh/Rician Fading
s
1 /
Flat/Selective Fading
Z
Slow/Flat fading
Summary
= e d
SISO Data Model 0
Channel Fluctuation
Error Rate
Outage Probability = 1 es /
Receive Diversity

Transmit Diversity
Again, as ( )
MIMO Information Theory
1
Course Summary Pout

END

Waveform Diversity and Design Conference, Slide #20

Introduction and Overview

Basic Digital Communications

So, what are we going to do about this?
Basic Wireless Communications
Fading Channels
Small Scale Fading
Rayleigh/Rician Fading
Flat/Selective Fading
Slow/Flat fading
Summary
SISO Data Model Use multiple antennas!
Channel Fluctuation
Error Rate
Outage Probability

Receive Diversity

Transmit Diversity

MIMO Information Theory

which provide diversity
Course Summary

END

Waveform Diversity and Design Conference, Slide #21

 Introduction to Receive Diversity
SIMO : Single antenna at the transmitter, multiple at the
receiver
Introduction and Overview
h
0
Basic Digital Communications Receiver 1

Basic Wireless Communications h

1
Receive Diversity
Receiver 2
Introduction
Diversity Types
Transmitter
Selection Combining
Maximal Ratio Combining h
N
Diversity Order Receiver N
Equal Gain Combining
Comparing Schemes
Summary
Correlation
If for a single receiver, Pout = 0.1, for two receivers
Transmit Diversity
Pout = 0.01
MIMO Information Theory
exponential gains in error rate with linear increase in
Course Summary number of antennas
END fundamental assumption: the error events are
independent, i.e, the channels are independent
Key: provide the receiver with multiple independent copies
of the message

Waveform Diversity and Design Conference, Slide #22

 Diversity Basics
Assume for now that the channels are independent and
identically distributed (i.i.d.)
Introduction and Overview
We will deal with the issue of correlation later
Basic Digital Communications

Basic Wireless Communications

Channel to nth receive element = hn , i.e., hn is assumed
Receive Diversity
independent of hm for n 6= m
Introduction
Diversity Types Signal to noise ratios are also i.i.d.: n is independent of m ,
Selection Combining
Maximal Ratio Combining n 6= m. Also,
Diversity Order
1
Equal Gain Combining
Comparing Schemes n en /
Summary
Correlation
Note: every channel has the same average SNR
Transmit Diversity

MIMO Information Theory

Course Summary

END

Waveform Diversity and Design Conference, Slide #23

 Basics (cont...)
x1

w1*
Introduction and Overview x2
Output Signal
Basic Digital Communications
w2*

Basic Wireless Communications
wN *
Receive Diversity
Introduction
xN
Diversity Types
Selection Combining
Maximal Ratio Combining
Diversity Order
Equal Gain Combining This appears to be beamforming! (for a single user!).
Comparing Schemes
Summary
Correlation Write the received signal as a vector
Transmit Diversity

MIMO Information Theory x = hs + n

T
Course Summary h = [h1 , h2 , . . . hN ]
END

The output signal is given by:

N
X
y= wn xn = wH x = wH hs + wH n
n=1

Waveform Diversity and Design Conference, Slide #24

 Receive Diversity Techniques
T
The weight vector is w = [w1 , w2 , . . . wN ]

Introduction and Overview

Key: how are these weights chosen?
Basic Digital Communications

Basic Wireless Communications

Receive Diversity Techniques:
Receive Diversity
Introduction Selection Combining
Diversity Types
Selection Combining
Maximal Ratio Combining Maximal Ratio Combining (MRC)
Diversity Order
Equal Gain Combining
Comparing Schemes
Equal Gain Combining (EGC)
Summary
Correlation

Transmit Diversity

MIMO Information Theory

Course Summary

END

Waveform Diversity and Design Conference, Slide #25

 Selection Combining
Choose, for further processing, the receive element with the
highest SNR
Introduction and Overview
(
Basic Digital Communications
1 k = maxn {n }
Basic Wireless Communications
wk =
0 otherwise
Receive Diversity
Introduction
Diversity Types
Selection Combining
The output SNR is therefore the maximum of the receive
Maximal Ratio Combining elements
Diversity Order
Equal Gain Combining
Comparing Schemes
Summary
Output SNR = out = max {n }
Correlation
n

Transmit Diversity

MIMO Information Theory

Note: channel phase information not required in the
Course Summary
selection process
END

This is the simplest diversity scheme

Seems to waste (N 1) receivers

Waveform Diversity and Design Conference, Slide #26

 Analyzing Selection Diversity
Outage probability : the output SNR is below threshold s if
all receive elements have SNR below s :
Introduction and Overview

Basic Digital Communications

Pout = P [out < s ]
Basic Wireless Communications = P [1 , 2 , . . . N < s ]
Receive Diversity N
Y
Introduction
Diversity Types = P [n < s ]
Selection Combining
Maximal Ratio Combining n=1
Diversity Order
Equal Gain Combining h iN
Comparing Schemes
Summary
Pout = 1 es /
Correlation

s /
 
Transmit Diversity
Pout = 1 e for N = 1, i.e.,exponential gains in
MIMO Information Theory
outage probability
Course Summary

END
Note: at high SNR, as
 N
1
Pout

Waveform Diversity and Design Conference, Slide #27

 Performance: Outage Probability versus SNR
0
10
N=1
Introduction and Overview
N=2
N=4
Basic Digital Communications
1
10
Basic Wireless Communications

Receive Diversity
Introduction
Outage Probability (P out)

Diversity Types
2
Selection Combining 10
Maximal Ratio Combining
Diversity Order
Equal Gain Combining
Comparing Schemes
Summary 3
10
Correlation

Transmit Diversity

MIMO Information Theory

4
10
Course Summary

END

0 5 10 15 20
SNR (dB)

In this figure, s = 0dB

Waveform Diversity and Design Conference, Slide #28

 Performance: Outage Probability versus s /
0
10

N=1
Introduction and Overview
N=2
N=3
Basic Digital Communications N =4

Basic Wireless Communications 1

10
Receive Diversity
Introduction
Fsc(s)

Diversity Types
Selection Combining

Maximal Ratio Combining

Diversity Order 2
Equal Gain Combining
10
Comparing Schemes
Summary
Correlation

Transmit Diversity
3
MIMO Information Theory 10

Course Summary

END

4
10
40 35 30 25 20 15 10 5 0 5 10
s/ (dB)

Waveform Diversity and Design Conference, Slide #29

 Performance: Bit Error Rate
0
10
SISO
Introduction and Overview
Selection with 4 branches

Basic Digital Communications

1
10
Basic Wireless Communications

Receive Diversity
Introduction
Diversity Types
2
Selection Combining 10
Maximal Ratio Combining
Diversity Order
BER

Equal Gain Combining

Comparing Schemes
Summary 3
10
Correlation

Transmit Diversity

MIMO Information Theory

4
10
Course Summary

END

0 2 4 6 8 10 12 14 16 18 20
= Average SNR (dB)

Waveform Diversity and Design Conference, Slide #30

 Analysis of Selection Combining (cont...)

Q: Are the gains in selection due to gains in SNR?

Introduction and Overview

Basic Digital Communications

A: In fact, no!!
Basic Wireless Communications

Receive Diversity
Introduction
Diversity Types
SNR analysis:
Selection Combining
Maximal Ratio Combining Note, Pout = P (out < s ) is also the cumulative density
Diversity Order
Equal Gain Combining function (CDF) of output SNR
Comparing Schemes
Summary
Correlation the probability density function, f (out ) = dPout /dout
Transmit Diversity

N out / h iN 1
out /
MIMO Information Theory
f (out ) = e 1e
Course Summary
END Z
Also, E {out } = out f (out )dout
0

Waveform Diversity and Design Conference, Slide #31

 SNR Analysis (cont...)

N
X 1
E {out } = ,
Introduction and Overview

n=1
n
Basic Digital Communications
 
Basic Wireless Communications
1
C + ln N + ,
Receive Diversity
Introduction 2N
Diversity Types
Selection Combining
Maximal Ratio Combining
The gain in SNR is only ln(N)!!!
Diversity Order
Equal Gain Combining
Comparing Schemes
Summary
Correlation

Transmit Diversity

MIMO Information Theory

Course Summary

END

Waveform Diversity and Design Conference, Slide #32

 SNR Analysis (cont...)

Q: So, where are the gains coming from?

Introduction and Overview
A: Reduced variation in the channel
Basic Digital Communications

Basic Wireless Communications

10
Receive Diversity
Introduction
Diversity Types 5
Selection Combining
Maximal Ratio Combining
Diversity Order 0
Equal Gain Combining
Channel Magnitude (dB)

Comparing Schemes
Summary
5
Correlation

Transmit Diversity
10
MIMO Information Theory

Course Summary 15

END

20

No diversity
4 branch selection
25
10 20 30 40 50 60 70 80 90 100
Sample #

Waveform Diversity and Design Conference, Slide #33

 Maximal Ratio Combining
Selection is simple, but wastes (N 1) receive elements

Introduction and Overview

Maximal Ratio Combining maximizes output SNR out
Basic Digital Communications
x = hs + n
h = [h1 , h2 , . . . hN ]T
Basic Wireless Communications

Receive Diversity
Introduction N
X
Diversity Types
Selection Combining Output Signal = y = wn xn = wH hs + wH n
Maximal Ratio Combining
Diversity Order n=1
Equal Gain Combining  H 2
Comparing Schemes w h E{|s|2 }
Summary
Correlation
Output SNR = out = n o
2
Transmit Diversity
E |wH n|
 H 2
MIMO Information Theory
w h E{|s|2 }
Course Summary =
END
2 ||w||2
"
H
2 #
w h
MRC: wMRC = max [out ] = max 2
w w ||w||2

Waveform Diversity and Design Conference, Slide #34

 Maximal Ratio Combing (cont...)
Using Cauchy-Schwarz inequality

Introduction and Overview

wh
Basic Digital Communications
Choose w = h,
Basic Wireless Communications
2 2 2


Receive Diversity Output Signal = y = |h1 | + |h2 | + . . . |hN | s + noise
Introduction
N
Diversity Types
!
Selection Combining
X
Maximal Ratio Combining = |hn |2 s + noise
Diversity Order
Equal Gain Combining n=1
Comparing Schemes
N N
Summary X E{|s|2 }|hn |2 X
Correlation
Output SNR = out = 2
= n
Transmit Diversity
n=1
n=1
MIMO Information Theory

Course Summary
i.e., the output SNR is the sum of the SNR over all receivers
END PDF of output SNR:
N 1
1 out out /
f (out ) = f (1 ) f (2 ) f (N ) = e ,
(N 1)! N

Waveform Diversity and Design Conference, Slide #35

 MRC Results
Average SNR
N
X
Introduction and Overview
out = n E{out } = N
Basic Digital Communications
n=1
Basic Wireless Communications

Receive Diversity Outage probability

Introduction
Diversity Types
N 1 
Selection Combining
s /
X s n 1
Maximal Ratio Combining
Diversity Order
Pout = P [out < s ] = 1 e
Equal Gain Combining
n=0
n!
Comparing Schemes
Summary
Correlation At high SNR ( )
Transmit Diversity
 N
MIMO Information Theory 1
Pout
Course Summary
END
Similarly, bit error rate:
Z  N
1
BER = [BER/out ] f (out )dout
0

Waveform Diversity and Design Conference, Slide #36

 Performance: Outage Probability
0
10
N=1
N=2
Introduction and Overview N=3
N =4
Basic Digital Communications

Basic Wireless Communications 1

10
Receive Diversity
Introduction
F (s)

Diversity Types
sc

Selection Combining
Maximal Ratio Combining
Diversity Order 2
Equal Gain Combining
10
Comparing Schemes
Summary
Correlation

Transmit Diversity
3
MIMO Information Theory 10

Course Summary

END

4
10
40 35 30 25 20 15 10 5 0 5 10
s/ (dB)

Outage probability versus s /

Waveform Diversity and Design Conference, Slide #37

 Performance: Bit Error Rate
0
10
SISO
Introduction and Overview
Selection with 4 branches
MRC with 4 branches
Basic Digital Communications
1
10
Basic Wireless Communications

Receive Diversity
Introduction
Diversity Types
2
Selection Combining 10
Maximal Ratio Combining
Diversity Order
BER

Equal Gain Combining

Comparing Schemes
Summary 3
10
Correlation

Transmit Diversity

MIMO Information Theory

4
10
Course Summary

END

0 2 4 6 8 10 12 14 16 18 20
= Average SNR (dB)

Waveform Diversity and Design Conference, Slide #38

 Diversity Order
A fundamental parameter of diversity-based systems

Introduction and Overview

Several times now we have seen that in the high-SNR regime
Basic Digital Communications
 N
1
Basic Wireless Communications BER or Pout

Receive Diversity
Introduction
log(BER)
Diversity Types
= N (at high SNR)
Selection Combining
Maximal Ratio Combining
log
Diversity Order
Equal Gain Combining
Comparing Schemes
Summary
Correlation
i.e., at high SNR the slope of the curve in a log-log plot is N
Transmit Diversity
this is the informal definition of diversity order
MIMO Information Theory simulations show that SNR need not be very high for this
Course Summary
to hold
END

Waveform Diversity and Design Conference, Slide #39

 Diversity Order (cont...)
Formal definition: The diversity order , D, is defined as

Introduction and Overview

log BER
Basic Digital Communications
D= lim
SNR log SNR
Basic Wireless Communications

Receive Diversity
Introduction
Diversity Types
Selection Combining
The diversity order measures the number of independent
Maximal Ratio Combining
Diversity Order
paths over which the data is received
Equal Gain Combining
Comparing Schemes
Summary
Can also use Pout in the definition
Correlation

Transmit Diversity
Diversity order is (formally) a high-SNR concept
MIMO Information Theory
Provides information of how useful incremental SNR is
Course Summary

END
Sometimes diversity order is abused:
the high-SNR definition masks system inefficiencies

Note: both selection and maximal ratio combining have

the same diversity order

Waveform Diversity and Design Conference, Slide #40

 The Gains are not due to SNR
0
10
N=1
N=2
Introduction and Overview N=4
No Fading
1
10
Basic Digital Communications

Basic Wireless Communications

2
Receive Diversity 10
Introduction
Diversity Types
Selection Combining
BER

3
Maximal Ratio Combining 10
Diversity Order
Equal Gain Combining
Comparing Schemes
Summary 4
10
Correlation

Transmit Diversity
5
10
MIMO Information Theory

Course Summary

6
END 10
10 5 0 5 10 15 20 25 30
N

BER versus output SNR. Note that even though output SNR is
the same, the BER is significantly different.

Waveform Diversity and Design Conference, Slide #41

 Equal Gain Combining
MRC requires matching of both phase and magnitude
Magnitude can fluctuate by 10s of dB
Introduction and Overview

Basic Digital Communications

Biggest gains are by the coherent addition
Basic Wireless Communications Equal Gain Combining: only cancel the phase of the channel
Receive Diversity
Introduction
Diversity Types
wn = ejhn
Selection Combining
Maximal Ratio Combining
Diversity Order
And so...
Equal Gain Combining
Comparing Schemes N
X
= wH x = wn xn
Summary
Correlation Output Signal = y
Transmit Diversity n=1
N
" #
MIMO Information Theory X
Course Summary = s |hn | + noise
END n=1

...resulting in a small loss in SNR...

h i
Average Output SNR = 1 + (N 1)
4

Waveform Diversity and Design Conference, Slide #42

 Comparing Diversity Schemes
Bit error rate:
SISO
Introduction and Overview 1
Selection with 4 branches
10
MRC
Basic Digital Communications EGC

Basic Wireless Communications

Receive Diversity
Introduction 2
10
Diversity Types
Selection Combining
Maximal Ratio Combining
BER

Diversity Order
Equal Gain Combining
Comparing Schemes
3
Summary 10
Correlation

Transmit Diversity

MIMO Information Theory

Course Summary 4
10

END

0 2 4 6 8 10 12 14 16 18 20
= Average SNR (dB)

Note: MRC and EGC have similar performance

Waveform Diversity and Design Conference, Slide #43

 Comparing Diversity Schemes (cont...)
Gains in SNR:
10
Introduction and Overview

9
Basic Digital Communications

Basic Wireless Communications 8

Receive Diversity
7
Introduction
Diversity Types
Selection Combining
Gain in SNR (dB)

6
Maximal Ratio Combining
Diversity Order
5
Equal Gain Combining
Comparing Schemes
Summary 4
Correlation

Transmit Diversity 3

MIMO Information Theory 2

Course Summary
1 Selection
END Maximal Ratio
Equal Gain
0
1 2 3 4 5 6 7 8 9 10
Number of elements (N)

Note: MRC and EGC have similar performance

Waveform Diversity and Design Conference, Slide #44

 Comparing Diversity Schemes (cont...)
All these diversity schemes have same diversity order
Work by reducing fluctuations in overall channel
Introduction and Overview
Selection Combining
Basic Digital Communications

Basic Wireless Communications

Simple to implement; only requires power measurement
Receive Diversity Gain in SNR = ln(N )
Introduction
Diversity Types
Selection Combining Maximal Ratio Combining
Maximal Ratio Combining
Diversity Order
Equal Gain Combining
Optimal in SNR sense
Comparing Schemes
Summary
Gain in SNR = N
Correlation

Transmit Diversity
Requires knowledge of channel and matching over
MIMO Information Theory
several 10s of dB
Course Summary
Easiest to analyze
END
Equal Gain Combining
Small loss w.r.t. MRC
Very difficult to analyze, but may be a good trade off for
implementation

Waveform Diversity and Design Conference, Slide #45

 In Summary...

Diversity is based on providing the receiver with multiple

Introduction and Overview independent copies of the same signal
Basic Digital Communications

Basic Wireless Communications

Receive Diversity
Introduction The key is the independence between the copies of the
Diversity Types
Selection Combining same signal
Maximal Ratio Combining
Diversity Order
Equal Gain Combining
The independence makes the gains in error rates
Comparing Schemes exponential with linear gains in number of elements
Summary
Correlation

Transmit Diversity
So, the question is ....under what circumstances can we
MIMO Information Theory assume independence?
Course Summary

END

Waveform Diversity and Design Conference, Slide #46

 The Issue of Correlation
Correlation between the received signals reduces the
independence and hence the effective diversity order
Introduction and Overview

Basic Digital Communications

The extreme case: if all elements were perfectly correlated
(e.g., line of sight conditions), diversity order = 1 (only SNR
Basic Wireless Communications
gains)
Receive Diversity
Introduction
Diversity Types
For two receive antennas, with correlation of :
Selection Combining
Maximal Ratio Combining 1 h out /(1+||) i
Diversity Order
f (out ) = e eout /(1||)
Equal Gain Combining
Comparing Schemes
2||
Summary
Correlation
1 h
Pout (s ) = 1 (1 + ||)es /(1+||)
Transmit Diversity
2||
MIMO Information Theory i
Course Summary (1 ||)es /(1||)
END

Correlation arises because

Electromagnetic Mutual Coupling
Finite distance between elements

Waveform Diversity and Design Conference, Slide #47

 Impact of correlation : Outage Probability
0
10

Introduction and Overview

Basic Digital Communications 1

10
Basic Wireless Communications

Receive Diversity
Introduction 2
Diversity Types 10
Selection Combining
F ( )
s

Maximal Ratio Combining

Diversity Order
Equal Gain Combining 3
Comparing Schemes
10
Summary
Correlation

Transmit Diversity
4
10
=0
MIMO Information Theory = 0.25
= 0.50
Course Summary = 0.75
=1
5
END 10
20 15 10 5 0 5 10
/ (dB)
s

Correlation below = 0.5 is considered negligible

Waveform Diversity and Design Conference, Slide #48

 Mutual Coupling

Introduction and Overview

Basic Digital Communications

Basic Wireless Communications

ZL ZL ZL
Receive Diversity
Introduction
Diversity Types
Selection Combining
Maximal Ratio Combining
Diversity Order
Equal Gain Combining
Comparing Schemes
Summary
Correlation
Voc = [Z + ZL ] Z1
L V
Transmit Diversity

MIMO Information Theory Z: A mutual impedance matrix

Course Summary

END
ZL : Diagonal load matrix
Voc : Open circuits voltages that would arise without mutual
coupling
V: True received voltages

Waveform Diversity and Design Conference, Slide #49

 Correlation due to Distance
If spacing between elements is d and signal arrives from
direction (, ) only, correlation is given by
Introduction and Overview

Basic Digital Communications

(, ) = ejkd cos sin ,
Basic Wireless Communications
where k = 2/
Receive Diversity
Introduction
Diversity Types
Total correlation is therefore averaged over angular power
Selection Combining
Maximal Ratio Combining
distribution
Diversity Order
Equal Gain Combining
Z Z 2
Comparing Schemes
Summary
= E{(, )} = ejkdcos sin f, (, )d d,
Correlation 0 0
Transmit Diversity
where f, (, ) is the power distribution of the received
MIMO Information Theory
signals over all angles
Course Summary

END
So, the angular distribution is crucial
Depends on where the receiver is
at the mobile or the base station
the base station looks down on the mobile

Waveform Diversity and Design Conference, Slide #50

 Correlation at the Mobile
The mobile is (usually) surrounded by many scatterers

Introduction and Overview

In a dense multipath environment
Basic Digital Communications
1
Basic Wireless Communications f, (, ) = ( 0 )
Receive Diversity Z2
1
( 0 )ejkdcos sin dd
Introduction
Diversity Types =
Selection Combining
, 2
Maximal Ratio Combining
Diversity Order
Equal Gain Combining
Comparing Schemes
= J0 (kd sin 0 )
Summary
Correlation

Transmit Diversity

MIMO Information Theory

0 = /2, < 0.5 if d > 0.24
Course Summary Required distance increases as 0 decreases
END
Rule of thumb: d /2
At 1GHz, = 30cm, i.e., received signals independent if
d > 15cm

Waveform Diversity and Design Conference, Slide #51

 Correlation at a Base Station

Introduction and Overview

Basic Digital Communications

Basic Wireless Communications Mobile

Receive Diversity R
Introduction
Diversity Types
Selection Combining
Maximal Ratio Combining
Diversity Order
D
Equal Gain Combining
Comparing Schemes
Summary

Correlation

Transmit Diversity

MIMO Information Theory

BS
Course Summary
Signal arrives from a small angular region surrounding the
END
mobile
Z max
= ejkd cos(+) sin 0 fB ()d,
max

Waveform Diversity and Design Conference, Slide #52

 Correlation at a Base Station (cont...)
Array along x-axis

Introduction and Overview

For a uniform disk of scatterers, = /2
Basic Digital Communications
2J1 (kd sin max sin 0 )
Basic Wireless Communications = .
Receive Diversity
kd sin max sin 0
Introduction
Diversity Types
Selection Combining
Maximal Ratio Combining R = 1.2km, D = 50m, 0 = 80o < 0.5 for d > 9
Diversity Order
Equal Gain Combining
Comparing Schemes
At 1GHz, received signals independent if d > 2.7m
Summary
Correlation
(approx. 9ft.)
Transmit Diversity

MIMO Information Theory The required distance is therefore determined by the array
Course Summary setting
END

Waveform Diversity and Design Conference, Slide #53

 Transmit Diversity
So far, we had a SIMO situation: a single transmitter and
multiple receivers
Introduction and Overview
Achieving diversity was relatively easy: each receiver
Basic Digital Communications
receives a copy of the transmitted signal
Basic Wireless Communications

Receive Diversity
multiple receivers multiple copies
Transmit Diversity
Transmit Diversity
What about the MISO situation?
Space-Time Coding
Alamoutis Scheme
Very useful in the expected asymmetrical communication
Design Criteria
Examples of Codes
scenarios with more traffic from base station to mobile
Multiplexing
Summary Base station is expensive, has more space, has multiple
MIMO Information Theory antennas
Course Summary Mobile is cheap, has little space, has one antenna
END
Users are downloading information, e.g., a webpage

Waveform Diversity and Design Conference, Slide #54

 Transmit Diversity (cont...)
MISO: N transmit antennas, one receive antenna

Introduction and Overview

Transmit diversity requires the time dimension. To see this,
Basic Digital Communications
consider if we did not use the time dimension. Each transmit
antenna transmits symbol s.
Basic Wireless Communications

Receive Diversity N
X
Transmit Diversity Received Signal = x = hn s + noise
Transmit Diversity
Space-Time Coding n=1
Alamoutis Scheme
Design Criteria
Examples of Codes = hs + n
Multiplexing
Summary

MIMO Information Theory N

X
Course Summary where h = hn
END n=1

Received signal is a scalar, there is no diversity here!!

And hence the concept of space-time coding

Waveform Diversity and Design Conference, Slide #55

 Space-Time Coding
Simple example: Transmit the same symbol over two time
slots (symbol periods)
Introduction and Overview
On time slot 1, antenna n = 1 transmits symbol s
Basic Digital Communications

Basic Wireless Communications

x1 = h1 s + n1
Receive Diversity

Transmit Diversity
On time slot 2, antenna n = 2 transmits the same symbol s
Transmit Diversity
Space-Time Coding
Alamoutis Scheme x2 = h2 s + n2
Design Criteria
Examples of Codes
Multiplexing
At the receiver form a receive vector over the two time slots
Summary
" # " # " #
MIMO Information Theory x1 h1 n1
x = = s+ = hs + n
Course Summary
x2 h2 n2
END

Maximum Ratio Combining:

H 2 2


y = h x = |h1 | + |h2 | s + noise
and we would get order-2 diversity

Waveform Diversity and Design Conference, Slide #56

 The (Famous) Alamoutis Code
The previous scheme wastes half the time

Introduction and Overview

A more efficient approach: consider two symbols s1 and s2
Basic Digital Communications In the first time slot,
Basic Wireless Communications antenna n = 1 transmits s1 and antenna n = 2 transmits s2
Receive Diversity

Transmit Diversity
x1 = h1 s1 + h2 s2 + n1
Transmit Diversity
Space-Time Coding
Alamoutis Scheme
In the second time slot,
Design Criteria
Examples of Codes
antenna n = 1 transmits s2 while antenna n = 2 transmits s1
Multiplexing
Summary
x2 = h1 s2 + h2 s1 + n2
MIMO Information Theory

Course Summary
One subtle, but important point: each element transmits with
END
half the available power
Form the received data vector (note the conjugate on x2 )
" # " #" # " #
x1 h1 h2 s1 n1
x= = +
x2 h2 h1 s2 n2

Waveform Diversity and Design Conference, Slide #57

 Alamouti (cont...)
Now work with x
" #
s1
Introduction and Overview
x = H +n
Basic Digital Communications s2
Basic Wireless Communications
" #
s1
Receive Diversity
y = HH x = HH H + noise
Transmit Diversity
s2
Transmit Diversity " #" #
|h1 |2 + |h2 |2
Space-Time Coding
Alamoutis Scheme 0 s1
Design Criteria = + noise
Examples of Codes 0 |h1 |2 + |h2 |2 s2
Multiplexing
Summary

MIMO Information Theory

i.e.,
2 2


Course Summary
y1 = |h1 | + |h2 | s1 + noise
END

2 2


y2 = |h1 | + |h2 | s2 + noise
we get order-2 diversity on both symbols!!

The key is that the effective channel matrix is orthogonal

Waveform Diversity and Design Conference, Slide #58

 Alamouti Code : Performance

SISO
1
10 Receive Diversity
Alamouti Scheme
Introduction and Overview

Basic Digital Communications

Basic Wireless Communications

2
Receive Diversity 10

Transmit Diversity
Transmit Diversity
BER

Space-Time Coding
Alamoutis Scheme
Design Criteria 3
Examples of Codes
10
Multiplexing
Summary

MIMO Information Theory

Course Summary 4
10
END

0 2 4 6 8 10 12 14 16 18 20
= Average SNR (dB)

Note: the diversity order of the Alamouti scheme and 2-branch

receive diversity is the same. Alamoutis scheme suffers a 3dB
loss because of the power splitting
Waveform Diversity and Design Conference, Slide #59
 Code Design Criteria: What Makes a Code Go
Alamoutis scheme is a space-time code
A careful organization of data (or a function of the data) in
Introduction and Overview
space and time
Basic Digital Communications

Basic Wireless Communications

Alamoutis scheme is specific to two transmit antennas...
Receive Diversity ...and, unfortunately, cannot be generalized to N > 2
Transmit Diversity
Transmit Diversity Consider a code with K symbols over N antennas and L
Space-Time Coding
Alamoutis Scheme time slots (code rate R = K/L)
Design Criteria
Examples of Codes
Multiplexing
Summary
c11 c12 c1L
c21 c22 c2L

MIMO Information Theory

Course Summary
C= .. .. .. ..
. . . .

END
cN 1 cN 2 cN L
here, cnl is a function of the K symbols transmitted in time
slot l using antenna n
e.g., in the Alamouti scheme, c12 = s2

Waveform Diversity and Design Conference, Slide #60

 Design Criteria (cont...)
We want to minimize error rate, the probability that codeword
C was transmitted and C was decoded
Introduction and Overview
   E 
s
P (C C) exp d2 C, C
Basic Digital Communications

Basic Wireless Communications 4 2

Receive Diversity

Transmit Diversity
Es is the available energy, = Es / 2
Transmit Diversity  
Space-Time Coding
Alamoutis Scheme
d C, C is the effective distance between C and C
Design Criteria
Examples of Codes  
Multiplexing
2
Summary d C, C = hH EEH h
MIMO Information Theory
Course Summary
c11 c11 c12 c12 c1L c1L
c21 c21 c22 c22 c2L c2L

END
E = .. .. .. ..
. . . .


cN 1 cN 1 cN 2 cN 2 cN L cN L
T
h = [h1 , h2 , . . . hN ] is the channel vector

Waveform Diversity and Design Conference, Slide #61

 Design Criteria (cont...)

c11 c11 c12 c12 c1L c1L
Introduction and Overview
c21 c21 c22 c22 c2L c2L


Basic Digital Communications E = .. .. .. ..
. . . .

Basic Wireless Communications
Receive Diversity cN 1 cN 1 cN 2 cN 2 cN L cN L
Transmit Diversity
Transmit Diversity
Space-Time Coding
Alamoutis Scheme
Design Criteria
Examples of Codes
Tarokh et al. developed two design criteria based on this
Multiplexing
Summary
error matrix:
MIMO Information Theory
The Rank Criterion: The maximum diversity order is
Course Summary achieved if the rank of the error matrix is maximized (N )
END If M receiving antennas, total diversity order available is
NM
The Determinant Criterion: The error rate is minimized if
the determinant of EEH is maximized over all code pairs
C, C

Waveform Diversity and Design Conference, Slide #62

 Examples of Space-Time Codes
Space-Time Trellis Codes: Design a trellis code over space
and time
Introduction and Overview

Basic Digital Communications

Orthogonal Block Codes: Codes in which each symbol can
be independently decoded
Basic Wireless Communications

Receive Diversity
Independent decoding is very convenient
Transmit Diversity
Transmit Diversity
Alamoutis code is orthogonal for N = 2
Space-Time Coding
Alamoutis Scheme
Rate-1 orthogonal codes cannot exist for N > 2
Design Criteria
Examples of Codes Rate-1/2 orthogonal codes are always available
Multiplexing
Summary
rate-3/4 codes are available for N = 3, 4, e.g.,
MIMO Information Theory

Course Summary
END
s1 s2 s3
s2 s1 s4
G3 = ,

s3 s4 s1
s4 s3 s2

Waveform Diversity and Design Conference, Slide #63

 Examples of Codes (cont...)
Linear Dispersion Codes: Minimize error based on mutual
information directly, not the design criteria
Introduction and Overview

Basic Digital Communications

Codes based on algebra: Algebraic codes, e.g., TAST etc.
Basic Wireless Communications In all cases, space-time codes provide diversity by giving the
Receive Diversity receiver independent copies of the same message
Transmit Diversity
Transmit Diversity
Space-Time Coding
Alamoutis Scheme
Design Criteria
Examples of Codes
Multiplexing
Summary
So far we have focused on diversity order (reliability) only.
MIMO Information Theory

Course Summary
What about data rate?
END

Waveform Diversity and Design Conference, Slide #64

 Space-Time Coding and Multiplexing
Transmit more than one data stream (multiplexing)
Requires multiple receive antennas as well
Introduction and Overview

Basic Digital Communications

Instead of transmitting only a single data stream, transmit Q
Basic Wireless Communications data streams in parallel.
Receive Diversity
M receive, N transmit antennas. Divide the transmit
Transmit Diversity
Transmit Diversity
antennas into Q groups, N = N1 + N2 + . . . NQ
Space-Time Coding
Alamoutis Scheme Data stream q uses Nq antennas
Design Criteria
Examples of Codes
Multiplexing
Summary
x = Hc + n,
MIMO Information Theory

Course Summary h11 h12 h1N1 h1(N1 +1) h1N c1
h21 h22 h2N1 h2(N1 +1) h2N c2
END


= .. .. .. .. .. .. ..
. . . . . . .


hM 1 hM 2 hM N1 hM (N1 +1) hM N cQ
+n

Waveform Diversity and Design Conference, Slide #65

 STC and Multiplexing (cont...)
The q th data stream uses a space-time code over Nq
antennas
Introduction and Overview

Basic Digital Communications

How do you isolate individual streams?

Basic Wireless Communications c2
Receive Diversity
c3

Transmit Diversity x = H1 c1 + H1 .. + n

Transmit Diversity
Space-Time Coding .
Alamoutis Scheme
Design Criteria cQ
Examples of Codes
Multiplexing
Summary Now, let H H
1 be the null space of H1 (H1 H1 = 0)
MIMO Information Theory
this is possible if M > N N1 . H
1 is size
Course Summary
M (M N + N1 )
END

H
 H 
y1 = H1 x = H1 H1 c1 + noise
which is a space-time coded system with N1 transmitters
and (M N + N1 ) receivers
Diversity order = N1 (M N + N1 )

Waveform Diversity and Design Conference, Slide #66

 Multiplexing (cont...)
Clearly we can apply the same idea for q = 2, . . . , Q
Data stream q would achieve diversity order of
Introduction and Overview
Nq (M N + Nq ).
Basic Digital Communications

Basic Wireless Communications

Can we do better? Yes!! Use interference cancellation...
Receive Diversity ...since c1 has been decoded, subtract it!
Transmit Diversity
Transmit Diversity
Data stream 2 sees less interference (N1 interfering
Space-Time Coding
Alamoutis Scheme
transmissions are eliminated). Stream 2 can get diversity
Design Criteria
Examples of Codes
order of N2 (M N + N2 + N1 )
Multiplexing
Summary Similarly, the q th data streamcan achieve diversity order
Pq
MIMO Information Theory
of Nq M N + p=1 Np
Course Summary

END BLAST: Bell Labs Layered Space-Time

Nq = 1

Requires M N (at least as many receivers as

transmitters)
Achieved (in lab) spectral efficiency of 10s of b/s/Hz!

Waveform Diversity and Design Conference, Slide #67

 In Summary...
Transmit diversity requires the time dimension
Space-time coding is the careful arrangement of data (or
Introduction and Overview
function of data) in space and time to achieve
Basic Digital Communications

Basic Wireless Communications

Greatest diversity order (rank criterion)
Receive Diversity Minimum error rate (determinant criterion)
Transmit Diversity
Transmit Diversity
Several space-time code families are available
Space-Time Coding
Alamoutis Scheme We focused on the simplest family of orthogonal
Design Criteria
Examples of Codes space-time block codes
Multiplexing
Summary
Can also use the spatial degrees of freedom to multiplex
MIMO Information Theory

Course Summary
BLAST is one (famous) example
END

Waveform Diversity and Design Conference, Slide #68

 MIMO Information Theory
We wish to investigate the fundamental limits of data transfer
rate in MIMO wireless systems
Introduction and Overview

Basic Digital Communications

Remember:
Basic Wireless Communications
A channel is fundamentally characterized by (and the data
Receive Diversity rate limited by) its capacity C
Transmit Diversity In the SISO case,
MIMO Information Theory
Basics
Unknown Channel
C = log2 (1 + SN R)
Known Channel
Waterfilling
MIMO Systems
Our MIMO system: N transmit and M receive antennas
Examples
Summary
DMT y = Hx + n
Course Summary

END
y: the length-M received signal vector
H: the M N channel
x: The length-N transmit data vector
Let Sx = E{xxH } be the covariance matrix of x

Waveform Diversity and Design Conference, Slide #69

 Channel Unknown at Transmitter
It is not too hard to show
 
1
Introduction and Overview C = log2 det I + 2 HSx HH
Basic Digital Communications

Basic Wireless Communications If channel H is known at the transmitter, Sx can be chosen to
Receive Diversity
best match the channel
Transmit Diversity

MIMO Information Theory

However, lets start with the case that the channel is not
Basics
Unknown Channel
known. The best choice is
Known Channel
Waterfilling Es
MIMO Systems Sx = IN N
Examples N
Summary  
1 Es
HHH
DMT
C = log2 det I + 2
Course Summary N
END
Since H is not known, again, one cannot guarantee a data
rate and the true capacity is zero!
Again, talk of an outage probability and/or expected
capacity

Waveform Diversity and Design Conference, Slide #70

 Unknown Channel (cont...)
 
1 Es H
C = log2 det I + HH
Introduction and Overview 2 N
Basic Digital Communications

Basic Wireless Communications

Since we are assuming Rayleigh fading, the entries of H are
Receive Diversity
complex Gaussian
Transmit Diversity Experts in STAP will recognize HHH as following the
MIMO Information Theory Wishart distribution
Basics
Unknown Channel
Known Channel
Let the eigenvalues of HHH be 2m , m = 1, 2, . . . , M
Waterfilling
MIMO Systems  
Examples Es
Eigenvalues of I + Es /(N 2 )HH H 2
 
Summary = 1+
DMT
(N 2 ) m
Course Summary

END

Waveform Diversity and Design Conference, Slide #71

 Unknown Channel (cont...)

M   M  
Y Es 2 X Es 2
C = log2 1+ 2
m = log2 1+ 2
m
Introduction and Overview

m=1
N m=1
N
Basic Digital Communications

Basic Wireless Communications

The distribution of m is known (HHH is Wishart)
Receive Diversity
Without ordering, these eigenvalues are independent and
Transmit Diversity
identically distributed
MIMO Information Theory
Basics
Unknown Channel
There are r = min(M, N ) eigenvalues
Known Channel
Waterfilling
MIMO Systems
Therefore, on average
Examples
M
Summary
(  )
DMT
X Es 2
E{C} = E{m } log2 1 + 2
m
Course Summary
m=1
N
END   
Es 2
= min(N, M )E log2 1 +
N 2
We get linear gains in capacity, not just power gains
As if we have min(N, M ) parallel channels!

Waveform Diversity and Design Conference, Slide #72

 Unknown Channel: Ergodic Capacity

Introduction and Overview

Basic Digital Communications

Basic Wireless Communications

Receive Diversity

Transmit Diversity

MIMO Information Theory

Basics
Unknown Channel
Known Channel
Waterfilling
MIMO Systems
Examples
Summary
DMT

Course Summary

END

Note: Ergodic capacity for fixed SNR (from Telatar (1999)).

Here r is the number of elements in the transmitter and
receiver (r = M = N )
Waveform Diversity and Design Conference, Slide #73
 Unknown Channel: 1-(Outage Probability)

Introduction and Overview

Basic Digital Communications

Basic Wireless Communications

Receive Diversity

Transmit Diversity

MIMO Information Theory

Basics
Unknown Channel
Known Channel
Waterfilling
MIMO Systems
Examples
Summary
DMT

Course Summary

END

Note: Success rates (probability that capacity is above target)

from Foschini and Gans (1998). Comparing success rates for
SISO and a N = M = 2 system.

Waveform Diversity and Design Conference, Slide #74

 Channel Known at the Transmitter
What if the channel is known at the transmitter?
Usually obtained via feedback (frequency division duplex -
Introduction and Overview
FDD) or via reciprocity (TDD)
Basic Digital Communications

Basic Wireless Communications

The transmitter can tune the covariance matrix Sx to
Receive Diversity
match the transmitter
Transmit Diversity This may be via power allocation
MIMO Information Theory
Basics
Unknown Channel Sopt
x = max C
Known Channel Sx
Waterfilling  
MIMO Systems 1 H
Examples = max log2 det I + HSx H
Summary Sx 2
DMT

Course Summary
Constraints: Sx must be positive-definite
END
Sx must satisfy a power constraint
So, how do you optimize over a matrix?
Lets start with a simpler (and very instructive) system:
parallel channels

Waveform Diversity and Design Conference, Slide #75

 Parallel Channels
1

2
Introduction and Overview

3
Basic Digital Communications

Basic Wireless Communications

Receive Diversity
N

Transmit Diversity

MIMO Information Theory

Basics
Each channel is independent of the other
Unknown Channel
Known Channel On the nth channel
Waterfilling
MIMO Systems
Examples yn = hn xn + noise
Summary
DMT

Course Summary
The transmitter has one important constraint - a total
END
available energy constraint of Es
Since the transmitter knows the channel values,
hn , n = 1, 2, . . . , N it can allocate power to maximize the
overall capacity

Waveform Diversity and Design Conference, Slide #76

 Parallel Channels (cont...)
The transmitter allocates power En to channel n
N  
2 En
Introduction and Overview
X
C= log2 1 + |hn | 2
Basic Digital Communications
n=1

Basic Wireless Communications

Receive Diversity Intuitively, the transmitter should allocate all its power to the
Transmit Diversity strongest channel, right?
MIMO Information Theory
Basics
Strangely enough, wrong!
Unknown Channel
Known Channel
Waterfilling
This is because...
MIMO Systems
Examples
Summary
C = log (1 + SNR)
DMT

Course Summary
At high SNR, C log (SNR)
END At low SNR, C SNR
There are diminishing marginal returns in allocating power

Waveform Diversity and Design Conference, Slide #77

 Parallel Channels (cont...)
The problem formulation:
N
En |hn |2
X  
Introduction and Overview  opt
En = max log2 1+
Basic Digital Communications {En }
n=1
n2
Basic Wireless Communications
N
X
Receive Diversity
En Es
Transmit Diversity
n=1
MIMO Information Theory
Basics En 0
Unknown Channel
Known Channel
Waterfilling The solution:
MIMO Systems
Examples
!
2
Summary
DMT
2 + En = , n = 1, . . . , N
Course Summary
|hn |
END

!+
2

En = 2 ,
|hn |

where (x)+ = 0 if x < 0 and (x)+ = x if x 0

Waveform Diversity and Design Conference, Slide #78

 Waterfilling
Note that is a constant

Introduction and Overview

Channel sees an effective noise variance of n2 = 2 /|hn |2
Basic Digital Communications
Power Level: Difference
Basic Wireless Communications

Receive Diversity

Transmit Diversity

MIMO Information Theory

Basics
Unknown Channel
Known Channel
Waterfilling
MIMO Systems
Examples
Summary
DMT

Course Summary
Channels 1 to N
END

Note that the better channels do get more power

Some channels are so bad that they do not get any power

Waveform Diversity and Design Conference, Slide #79

 MIMO Systems with Known Channel
So far, we have focused on parallel channels.

Introduction and Overview h1 0 0
0 h2 0
Basic Digital Communications

Basic Wireless Communications
So far, y = .. .. . . x + n
..
. . . .
Receive Diversity

Transmit Diversity
0 0 hN
MIMO Information Theory
Basics What does this tell us about a regular MIMO system?
Unknown Channel
Known Channel
Waterfilling We have
MIMO Systems
Examples y = Hx + n
Summary
DMT N transmitters, M receivers, H is the M N channel
Course Summary

END
h11 h12 h1N
h21 h22 h2N


H= .. .. .. ..
. . . .


hM 1 hM 2 hM N

Waveform Diversity and Design Conference, Slide #80

 MIMO Systems (cont...)
One can use the singular value decomposition of H

Introduction and Overview

H = UVH
Basic Digital Communications
U is the matrix of eigenvectors of HHH
Basic Wireless Communications

Receive Diversity V is the matrix of eigenvectors of HH H

Transmit Diversity
is a diagonal M N matrix of singular values
MIMO Information Theory
Basics
Unknown Channel
Known Channel
1 0 0 0 0
Waterfilling
0 2 0 0 0

MIMO Systems

. .
.. . . . . .. .
Examples

. .. . ..

Summary
.
DMT = ,
Course Summary
0
0 R 0 0
0 0 0 0 0
END

0 0 0 0 0
R is the rank of H
The zeros pad the matrix to match the M N dimensions

Waveform Diversity and Design Conference, Slide #81

 MIMO Systems (cont...)
How does this help? We have

Introduction and Overview UUH = UH U = IM M

Basic Digital Communications
VVH = VH V = IN N
Basic Wireless Communications

Receive Diversity

Transmit Diversity y = Hx + n
MIMO Information Theory
Basics
Unknown Channel
Known Channel
= UVH x + n
Waterfilling
MIMO Systems
Examples
Summary UH y = VH x + UH n
DMT

Course Summary

END
y = UH y, x = VH x, n = UH n

y = x + n

We have a set of R parallel channels!

Waveform Diversity and Design Conference, Slide #82

 MIMO Systems (cont...)
Therefore,
N outputs M inputs
Introduction and Overview
R outputs Routputs

Basic Digital Communications

data Power
S/P H data
Allocation V Channel U Decoder
Basic Wireless Communications H

Receive Diversity N-R zeros

~ ~
Transmit Diversity
x x y y
MIMO Information Theory
Basics
Unknown Channel
The transmitter precodes the transmitted signal using matrix
Known Channel V. This matches the transmission to the eigen-modes of
Waterfilling
MIMO Systems the channel
Examples
Summary
DMT
The transmitter also waterfills over the singular values n as
Course Summary the equivalent channel values as parallel channels (called hn
END earlier)
The receiver decodes using the matrix U
IMPORTANT: This diagonalization process (transmission on
eigen-modes is a fundamental concept in wireless
communications
Waveform Diversity and Design Conference, Slide #83
 Improvement in Capacity
12

Introduction and Overview

11
Basic Digital Communications

Basic Wireless Communications

10

Receive Diversity
9
Capacity (bits/channel use)

Transmit Diversity

MIMO Information Theory 8

Basics
Unknown Channel
Known Channel 7
Waterfilling
MIMO Systems
Examples 6
Summary
DMT

5
Course Summary

END
4
Without Waterfilling
With Waterfilling
3
0 2 4 6 8 10 12 14 16 18 20
= Average SNR (dB)

Comparing capacities for N = M = 4

Waveform Diversity and Design Conference, Slide #84

 Some Illustrative Examples
Example 1: SIMO System, 1 transmitter, M receivers

Introduction and Overview h1
h2
Basic Digital Communications

Basic Wireless Communications
H=h = ..

Receive Diversity
.
Transmit Diversity
hM
 
MIMO Information Theory
h
Basics
U = h
Unknown Channel
Known Channel
||h||
Waterfilling
MIMO Systems V = [1]
Examples
T
Summary
DMT
= [||h||, 0, . . . , 0]
Course Summary
Note that we have only a single parallel channel
END
 
Es 2
C = log2 1 + ||h||
2
Effectively, all channel powers added together

Waveform Diversity and Design Conference, Slide #85

 Some Illustrative Examples (cont...)
Example 2: MISO System, N transmitters, 1 receiver

Introduction and Overview

H = h = [h1 , h2 , . . . , hN ]
Basic Digital Communications U = [1]
 
Basic Wireless Communications
h
V = h
Receive Diversity
||h||
Transmit Diversity
U = [1]
MIMO Information Theory
Basics
Unknown Channel
= [||h||, 0, . . . , 0]
Known Channel
Waterfilling
MIMO Systems
Again we have only a single parallel channel
Examples
 
Summary
Es
DMT
C = log2 1 + 2 ||h||2
Course Summary
END
This is the same as the SIMO case!
Note: without channel knowledge,
 
Es 2
C = log2 1 + ||h||
N 2

Waveform Diversity and Design Conference, Slide #86

 Some Illustrative Examples (cont...)
Example 3: MIMO System, line of sight conditions
2 M 1 T
zr = ejkdr cos r
 
Introduction and Overview s(r ) = 1, zr , zr , . . . , zr ,
Basic Digital Communications
2 N 1 T
zt = ejkdt cos t
 
Basic Wireless Communications
s(t ) = 1, zt , zt , . . . , zt ,
Receive Diversity

Transmit Diversity H = s(r ) sT (t )

ztN 1

zt2
MIMO Information Theory
Basics 1 zt
Unknown Channel

zr zr zt zr zt2 zr ztN 1
Known Channel


Waterfilling
MIMO Systems
=
.. .. .. .. ..
. . . . .

Examples
Summary
DMT
zrM 1 zrM 1 zt zrM 1 zt2 zrM 1 ztN 1
Course Summary

END
This is a rank-1 matrix!

This one singular value = 1 = N M
   
Es Es
C = log2 1 + N M = log2 1 + 2 M
N 2

Waveform Diversity and Design Conference, Slide #87

 Some Illustrative Examples (cont...)
Example 4: MIMO System, M = N , rich scattering
conditions, full rank channel
Introduction and Overview
We have N parallel channels
Basic Digital Communications

Basic Wireless Communications

For convenience, assume all singular values are equal
Receive Diversity Let this singular value = 1
Transmit Diversity
Since all parallel channels are equally powerful, power
MIMO Information Theory
Basics allocation is uniform
Unknown Channel
Known Channel
Waterfilling
MIMO Systems N    
Examples
X Es 2 Es 2
Summary C= log2 1 + 2
1 = N log2 1 + 2
1
DMT
n=1
N N
Course Summary

END
Note the huge difference from line of sight scenario
The number of transmit or receive elements is outside the
log term; we get linear gains in capacity
 
Es 2
If N 6= M C = min(N, M ) log2 1 +
N 2 1

Waveform Diversity and Design Conference, Slide #88

 Summary of Information Theoretic Analysis
MIMO systems allow for huge increases in capacity
If the fading is independent then one can achieve linear
Introduction and Overview
gains in capacity over the SISO case
Basic Digital Communications

Basic Wireless Communications

Notice that this inherently requires the concept of a
Receive Diversity
diversity of paths
Transmit Diversity The concept of diagonalization or transmission on
MIMO Information Theory
Basics
eigen-channels and the associated concept of waterfilling
Unknown Channel are fundamental
Known Channel
Waterfilling
MIMO Systems Waterfilling allocates more power to better channels
Examples
Summary
DMT
Note that this is, initially, counter-intuitive. Generally, if we
Course Summary
have a poor channel, we add power, not reduce power
END

Waveform Diversity and Design Conference, Slide #89

 Diversity-Multiplexing Tradeoff
Consider a system with N transmit and M receive antennas in
a rich scattering, Rayleigh fading environment
Introduction and Overview
Diversity: We have seen that through space-time coding and
Basic Digital Communications
receive diversity we can achieve a diversity order of N M .
Basic Wireless Communications

Receive Diversity
Multiplexing: In the information theoretic analysis we saw
Transmit Diversity that we could get a pre-log factor of min(M, N ). Also, at high
MIMO Information Theory
SNR
Basics
Unknown Channel
C min(M, N ) log2 (SNR)
Known Channel
Waterfilling
MIMO Systems
Examples
Summary
DMT

Course Summary

END
Q: Can we get both diversity and multiplexing (rate) gains?
A: Yes! But, there is a trade-off between the two!

Waveform Diversity and Design Conference, Slide #90

 DMT (cont...)
As before, define the diversity order as:
 
Introduction and Overview log Pout
D= lim
Basic Digital Communications
SNR log SNR
Basic Wireless Communications

Receive Diversity
D tells us how fast the error rate falls with increases in
Transmit Diversity
log(SNR)
MIMO Information Theory
Basics Define a multiplexing gain r as
Unknown Channel
Known Channel
 
Waterfilling
MIMO Systems
R
Examples
r= lim
Summary SNR log SNR
DMT

Course Summary R is the rate of transmission

END
r is the rate at which the transmission rate increases with
log(SNR)

Waveform Diversity and Design Conference, Slide #91

 DMT (cont...)
Diversity Multiplexing Tradeoff: The optimal tradeoff curve,
d (r) is given by the piecewise-linear function connecting the
Introduction and Overview points (r, d (r)), r = 0, 1, . . . , min(M, N ), where
Basic Digital Communications

Basic Wireless Communications d (r) = (M r)(N r)

Receive Diversity

Transmit Diversity Note: dmax = M N and rmax = min(M, N ).

MIMO Information Theory
Basics At integer points, r degrees of freedom are used for
Unknown Channel
Known Channel multiplexing, the rest are available for diversity
Waterfilling
MIMO Systems
Examples Note the linear interpolation
Summary between integer points
DMT

Course Summary

END

Waveform Diversity and Design Conference, Slide #92

 Course Summary
We have explored uses of the spatial dimension in wireless
communications
Introduction and Overview

Basic Digital Communications

Performance of wireless systems is fundamentally limited by
fading
Basic Wireless Communications

Receive Diversity
Fading makes the received signal to be a random copy of
Transmit Diversity the transmitted signal
MIMO Information Theory MIMO systems are based on independent fading to/from
Course Summary multiple elements
Course Summary
Whats Left?
We covered three major concepts:
END

Receive Diversity
Transmit Diversity
MIMO Information Theory
Receive Diversity:
Selection, maximal ratio and equal gain combining
Trade off between complexity and performance

Waveform Diversity and Design Conference, Slide #93

 Course Summary (cont...)
Receive Diversity (cont...)
Notion of diversity order
Introduction and Overview

Basic Digital Communications

Measures the number of independent paths the signal is
Basic Wireless Communications
received
Receive Diversity Impact of correlation
Transmit Diversity Array elements must be some minimum distance apart
MIMO Information Theory
The key is to create independence
Course Summary
Course Summary
Whats Left? Transmit Diversity:
END Requires the time dimension: space-time coding
An important example of block codes: Alamoutis Code
Unfortunately, only rate 1/2 codes are guaranteed for
complex data constellations
Good codes designed using the rank and determinant
criteria

Waveform Diversity and Design Conference, Slide #94

 Course Summary (cont...)
Multiplexing:
Transmitting multiple data streams simultaneously
Introduction and Overview

Basic Digital Communications

Can combine space-time coding with multiplexing
Basic Wireless Communications A famous example: the BLAST scheme
Receive Diversity We should emphasize that this is only one possible
Transmit Diversity
multiplexing scheme
MIMO Information Theory

Course Summary
MIMO Information Theory:
Course Summary
Whats Left? Capacity via determinant of channel
END Ergodic capacity and outage probability if channel is
unknown
Transmission on eigen-channels and using waterfilling if
channel is known to transmitter
Creates r = rank (H) parallel channels
There is a fundamental trade-off between diversity and
multiplexing

Waveform Diversity and Design Conference, Slide #95

 We could have done so much more...
Channel Estimation: Throughout we assumed the receiver
knows the channel
Introduction and Overview
The receiver has to estimate and track a time varying
Basic Digital Communications
channel
Basic Wireless Communications

Receive Diversity
Frequency selective channels: We focused on flat channels,
Transmit Diversity frequency selectivity is becoming important
MIMO Information Theory Everyone assumes 4G will be based on OFDM -
Course Summary MIMO-OFDM is a hot topic
Course Summary
Whats Left?
Error control coding to achieve the capacity of MIMO
END
systems
Feedback to inform the transmitter of the channel
Low data rate schemes, error bounds, impact of error

Waveform Diversity and Design Conference, Slide #96

 What else? (cont...)
Multiuser Communications: Transmitting to/receiving from
multiple users simultaneously
Introduction and Overview
Is another form of multiplexing; requires interference
Basic Digital Communications
cancellation, provides flexibility
Basic Wireless Communications

Receive Diversity
Is getting more important, both theoretically and in
Transmit Diversity
implementation
MIMO Information Theory Cooperative Communications:
Course Summary Probably the hottest research area now
Course Summary
Whats Left?
Nodes with a single antenna share resources to act like a
END
MIMO system
via relaying, forwarding, cooperative diversity
Especially applicable to the new modern kinds of
networks. Also, distributed signal processing.
Sensor Networks: Large scale networks of small, cheap

nodes
Mesh Networks: Networks of access points

Waveform Diversity and Design Conference, Slide #97

 END
That all folks!
Introduction and Overview

Basic Digital Communications

Basic Wireless Communications

Receive Diversity

Transmit Diversity Again, detailed notes available at

MIMO Information Theory

Course Summary

END

http://www.comm.utoronto.ca/ rsadve/teaching.html

Waveform Diversity and Design Conference, Slide #98