MICAS922 - Part 1 MIMO

Ghaya Rekaya-Ben Othman

Plan

I Part I: Fading Channel Model

I Part II: MIMO System and Space-Time Coding

I Part III: Decoding chain for MIMO system

I Part IV: MIMO Precoding

 Part I

Fading Channel Model

Part I : Fading Channel Model

Fading Channels

Diversity Techniques

Information Theory Results

Example of constellation diversity

Fading Channel Model

Fading Channels

Diversity Techniques

Information Theory Results

Example of constellation diversity

Mobile radio channel model MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
I A mobile-radio channel is characterized by fading transmission system

Diversity Techniques
Concept of Diversity
Diversity Techniques

Information Theory
Results
Ergodic capacity
Outage Probability

Example of constellation
I In the presence of a large number of fading, the mobile radio channel diversity
can be modeled by a complex Gaussian random process (using central Choice of the constellation
function of the channel
limit theorem). Construction Criteria

I If the average of the complex envelope is null: it’s called a Rayleigh

channel

I Otherwise: it’s a Rice Channel.

Rayleigh channel MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
transmission system

I Ergodic channel (Fast-Fading channel): a realization of the channel Diversity Techniques

Concept of Diversity
is seen at each symbol time. Diversity Techniques

Information Theory
I Quasi-static fading channel: the channel remains constant during Results
the transmission of a frame or a code word. Ergodic capacity
Outage Probability

I Block-fading channel: the channel remains constant during the Example of constellation
diversity
transmission of n frames. If n = 1, the channel is quasi-static. Choice of the constellation
function of the channel
Construction Criteria
To the transmission channel is added AWGN noise (radiation picked up by
the antennas, possible interference between users and noise generated by
electronic components).
Point-to-point link MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

The received signal is written : Fading Channels

Mobile Radio Channel
Model of a mobile radio
y = α ·x +b transmission system

Diversity Techniques
Concept of Diversity

I b : the additive white Gaussian noise Diversity Techniques

Information Theory
I x : the information symbol taken from a given complex alphabet Results
Ergodic capacity
I α: fading coefficient, following a Gaussian distribution of zero mean
Outage Probability
and 0.5 variance per component.
Example of constellation
diversity
Choice of the constellation
Vectorial system Model : function of the channel
Construction Criteria

y = x·α +b

 
α1 ··· 0
 . .. . 
[y1 · · · yn ] = [x1 · · · xn ] ·  . . .  + [b1 · · · bn ]
 . . 
0 ··· αn
General Model MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
transmission system
The general model of a mobile radio transmission system is : Diversity Techniques
Concept of Diversity
y = x·H+b Diversity Techniques

Information Theory
Results
H represents the mobile radio channel. Ergodic capacity
Outage Probability

Example of constellation
I MIMO system: the coefficients of the H matrix correspond to the diversity
fading between the different antennas during transmission and Choice of the constellation
function of the channel
reception. Construction Criteria

I OFDM system: the coefficients of the matrix H correspond to the

fading of each carrier.
Fading Channel Model

Fading Channels

Diversity Techniques

Information Theory Results

Example of constellation diversity

Diversity: definition MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

I In the presence of deep fading, sending a single replica of the signal Fading Channels
Mobile Radio Channel
may not be enough to decode the information.
Model of a mobile radio
transmission system
I The solution would be to have different versions of the signal with Diversity Techniques
independent fading on reception. Concept of Diversity
Diversity Techniques

Definition Information Theory

Results
Diversity of order n is the reception of the same signal on n independent Ergodic capacity
channels. Outage Probability

Example of constellation
diversity
I The Contribution of Diversity Choice of the constellation
function of the channel
I For a Rayleigh channel, the error probability is asymptotically
Construction Criteria
proportional to the signal to noise ratio.γ
1
Pe '
γ

I By having a diversity of order n, the probability of error becomes :

 n
1
Pe '
γ
Diversity Techniques (1) MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
transmission system

There are two types of diversity techniques: Diversity Techniques

Concept of Diversity
Diversity Techniques

I The first type: does not require any repetition or coding Information Theory
Results
I Space Diversity in reception: Consists of receiving the same signal on n
Ergodic capacity
different antennas separated by at least 10 times the wavelength. The Outage Probability
maximum possible diversity order is equal to n.
Example of constellation
diversity
I Path Diversity: Consists of receiving n replicas of the signal from Choice of the constellation
function of the channel
different paths. A Rake receiver allows to decorrelate the different
Construction Criteria
paths. The maximum diversity order is equal to the number of paths
that can be decorrelated. This diversity technique is used in DS-CDMA
systems.
Diversity Techniques (2) MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
transmission system
I The second type: requires repetition and/or coding Diversity Techniques
Concept of Diversity
I Frequency diversity: consists in sending the same signal on different Diversity Techniques

frequencies separated by at least the channel coherence band. This Information Theory
diversity technique is used in OFDM systems. Results
Ergodic capacity
Outage Probability
I Temporal diversity: Consists of sending the same signal at different
Example of constellation
times separated by at least the channel coherence time to ensure their diversity
decorrelation. This diversity technique is interesting for ergodic Choice of the constellation
channels. function of the channel
Construction Criteria

I Space diversity in transmission: consists in sending the same signal on

different antennas separated by at least 10 times the wavelength. At
reception, this diversity is perceived as temporal diversity.
Diversity with coding MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
I Corresponds only to the second type of diversity techniques. transmission system

Diversity Techniques
I As they are defined, diversity techniques of the second type,
Concept of Diversity
correspond to the use of a repetition code (n, 1), which leads to a Diversity Techniques
waste of resources.
Information Theory
Results
I The solution can be to replace the repetition code with a code (n, k) Ergodic capacity
with minimum distance dmin . If the order of physical diversity is greater Outage Probability

than dmin , then the error probability becomes : Example of constellation

diversity
1
Pe ∼ Choice of the constellation
Γdmin function of the channel
Construction Criteria
I Let ∆ be the physical diversity order of the system. The maximum
order of diversity that can be obtained by using channel coding is:
min (dmin , ∆)
Combination of diversity techniques MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
I In practical systems, in order to achieve high orders of diversity as well Model of a mobile radio
transmission system
as high spectral efficiencies, several diversity techniques are combined.
Diversity Techniques
I The total order of diversity is equal to the product of the different Concept of Diversity
orders of diversity. Diversity Techniques

Information Theory
Results
I The most commonly used combinations are : Ergodic capacity
I Time-Frequency Outage Probability

I Space-Time Example of constellation

I Space-Frequency diversity
I Path-Space Choice of the constellation
I function of the channel
Path-Time
Construction Criteria

I For MIMO systems with nt transmit antennas and nr receive

antennas, the maximum diversity order that can be obtained is nt · nr .

How do to achieve this diversity order ?

Fading Channel Model

Fading Channels

Diversity Techniques

Information Theory Results

Example of constellation diversity

Mutuel Information and capacity MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman
How much information will the source be able to transmit on a MIMO
channel? Fading Channels
Mobile Radio Channel

I Let’s suppose that the inputs X are Gaussian. Model of a mobile radio
transmission system

I Note Q the covariance matrix of the Gaussian input X, Diversity Techniques

Concept of Diversity
h i
Diversity Techniques
Q = E X† X .
Information Theory
Results
The mutual information is : Ergodic capacity

  Outage Probability
1
I (X; Y) = log2 det Inr + 2 HQH† Example of constellation
diversity
σ
Choice of the constellation
function of the channel
Construction Criteria
I The channel capacity is defined as the maximum mutual information
on all possible distributions of the input X.

To reach the channel capacity, it is necessary to find

 
1
Q̂ = arg max log2 det Inr + 2 HQH†
Q σ

under the constraint that the total emitted energy is limited to PT .

Ergodic capacity MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
I The channel varies fast enough to confuse temporal average and Model of a mobile radio
statistical average. transmission system

Diversity Techniques

I Choose Q = PT Int is optimal, which corresponds to the decorrelated Concept of Diversity

nt Diversity Techniques
case and of the same power. Information Theory
Results
Ergodic capacity
I In this case, one can calculate the ergodic capacity Outage Probability

     Example of constellation
SNR diversity
C (SNR) = EH max I (X; Y) = EH log2 det Inr + HH† Choice of the constellation
Q≤PT nt function of the channel
Construction Criteria
= min {nt , nr } log2 SNR + O(1)

where min {nt , nr } is the number of degrees of freedom of the channel.

I The capacity of the ergodic channel corresponds to the capacity of

min {nt , nr } parallel AWGN channels.
Outage Probability MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
I For a block fading channel, the capacity in Shannon’s sense is always Othman
zero.
Fading Channels
I The maximized instantaneous capacity is: Mobile Radio Channel
Model of a mobile radio
transmission system
 
SNR
C (H) = log2 det Inr + HH† Diversity Techniques
nt
Concept of Diversity
Diversity Techniques

Information Theory
I C (H) is considered a random variable function of the instantaneous Results
channel response, which is constant during the transmission of a code Ergodic capacity
Outage Probability
word of finite length.
Example of constellation
I Outage Probability is : diversity
Choice of the constellation
function of the channel

Pout = Pr {C (H) ≤ R} Construction Criteria

where R is the spectral efficiency in bits per channel use.

Asymptotic behaviour
at high signal-to-noise ratio, we have:
 
1
Pout = O
SNRd

where d is the system physical diversity order (for example, nt · nr ).

Fading Channel Model

Fading Channels

Diversity Techniques

Information Theory Results

Example of constellation diversity

Constellation choice MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
We consider a SISO (nt = nr = 1 ) transmission, on a fast fading channel transmission system
(αis varying each time slot) : Diversity Techniques
Concept of Diversity
Diversity Techniques
y = α ·x +n
Information Theory
Results
Ergodic capacity
The questions that arise : Outage Probability

Example of constellation
I How to choose good constellations? diversity
Choice of the constellation
function of the channel
Construction Criteria
I On which parameters the choice should be based on?

I Are constellations that are good for the AWGN channel also good for
the Fast Fading channel?
q − QAM Constellation MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
I q − QAM constellation are finite sub-sets of Z[i]. Model of a mobile radio
transmission system

Diversity Techniques
Concept of Diversity
Diversity Techniques

Information Theory
Results
Ergodic capacity
Outage Probability

Example of constellation
diversity
Choice of the constellation
function of the channel

Canal AWGN Canal de Rayleigh Construction Criteria

(α1 , α2 ) = (1, 0.5)

I q − QAM constellations are adapted for AWGN channel, but not for
Fast Fading channel.
Turned q − QAM constelaltion MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
I A possible solution to adapt q − QAM constellations to the Rayleigh Model of a mobile radio
transmission system
channel, would be to rotate them.
Diversity Techniques
Concept of Diversity
Diversity Techniques

Information Theory
Results
Ergodic capacity
Outage Probability

Example of constellation
diversity
Choice of the constellation
function of the channel
Construction Criteria
Canal AWGN Canal de Rayleigh
(α1 , α2 ) = (1, 0.5)

I The new turned constellation is more resistant to fainting.

Error probability derivation MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

I In order to determine the criteria for the construction of good Fading Channels
Mobile Radio Channel
constellations, we will derive an upper bound of the error probability. Model of a mobile radio
transmission system

I q − QAM constellation are finite sub-sets of Z[i]. Since Z[i] is Diversity Techniques
geometrically uniform, we can then define Pe (Z[i]) instead of Concept of Diversity
Diversity Techniques
Pe (q − QAM).
Information Theory
Results
I Let S be a constellation defined as a subset of a lattice Λ, we then Ergodic capacity
Outage Probability
have :
Pe (S) ≤ Pe (Λ) Example of constellation
diversity
Choice of the constellation
function of the channel
Construction Criteria
I By using the union bound, we obtain an upper bound of Pe

Pe (Λ) ≤ ∑ P(x → y )
y 6=x

where P(x → y ) is the pairwise error probability, the probability to decide

y , while x has been transmitted. y and xare considered the unique
elements of the constellation.
Error probabilty bound for AWGN channel MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
transmission system

I For AWGN channel, we have : Diversity Techniques

Concept of Diversity
 
dE ,min Diversity Techniques
Pe (Λ) ≤ τQ √ Information Theory
2 N0 Results
Ergodic capacity
Outage Probability
where τ is the kissing number ( the average number of closest neighbors )
and dE ,min the minimum euclidean distance of the constellation. Example of constellation
diversity
Choice of the constellation
I The error probability bound is dominated by dE ,min . function of the channel
Construction Criteria

I A good constellation for an AWGN channel admits a relatively high

dE ,min .
Error probabilty bound for Rayleigh channel MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
Model of a mobile radio
transmission system
I For Fast Fading Rayleigh channel, using either Chernoff’s inequality
Diversity Techniques
or an approximation of the function Q(x) :
Concept of Diversity
Diversity Techniques

n Information Theory
1 1 Results
Pe (x → y ) ≤ ∏ (xi −yi )2
2 i=1 1+ Ergodic capacity
8N0 Outage Probability

Example of constellation
diversity
I At high signal-to-noise ratio : Choice of the constellation
function of the channel
Construction Criteria

1 n 1
Pe (x → y ) ≤ ∏ (xi −yi )2
2 i=1
8N0
Product distance MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

I Let the product distance produit be: Fading Channels

Mobile Radio Channel

(l) ∏xi 6=yi (xi − yi )2 Model of a mobile radio

dp (x, y )2 = transmission system

(E /n)l Diversity Techniques

Concept of Diversity
Diversity Techniques
where l represents the Hamming distance between x and y , and Information Theory
E = E (kxk2 ) the average symbole energy of the constellation. Results
Ergodic capacity
Outage Probability
I The pairwise error probability bound is can be written as a function
Example of constellation
of the product distance: diversity
Choice of the constellation
function of the channel
1 1
Pe (x → y ) ≤ l Construction Criteria
2

E (l)
8nN0 dp (x, y )2

I Let L = minx6=y (l), the error probability of error is dominated by the

1

L
term RSB , we call L the diversity order of the constellation.

I if L = n, the constellation has maximum diversity.

Constellation diversity MICAS922 - Part 1
MIMO

Ghaya Rekaya-Ben
Othman

Fading Channels
Mobile Radio Channel
I If two constellations have the same diversity order, we compare their Model of a mobile radio
(L) transmission system
dp,min .
Diversity Techniques
Concept of Diversity
Diversity Techniques
I To build a good constellation for a Rayleigh channel you need to
Information Theory
Results
I First: Maximize the order of diversity Ergodic capacity
Outage Probability
I Second: Maximize the minimum product distance
Example of constellation
diversity
Choice of the constellation
function of the channel
Construction Criteria

I The QAM constellation has an order of diversity equal to 1, while the

turned QAM constellation has an order of diversity equal to 2.

I The best rotation is the one that maximizes the minimum product
distance.