Modern Wireless Networks

Wireless Fundamentals
ICEN 574– Spring 2019
Prof. Dola Saha
1
 Wireless Digital Communication System
Bits Symbols Digital Waveform Digital / RF
Encoder
Modulator Analog Module

Wireless
Channel

Bits De- Analog / RF

Decoder
modulator Digital Module
2
Multipath Channel Effects
Ø x

3
Wideband vs Narrowband

4
Effect of dispersion (Inter Symbol Interference)

5
ISI as an impediment to increase data rate
Ø Need for higher data rate urges to transmit at higher
symbol è Higher ISI

6
Multicarrier Modulation to combat ISI
Ø Divides the wideband incoming data stream into L
narrowband substreams
Ø Each substream is then transmitted over a different
orthogonal frequency subchannel
Ø Number of substreams L is chosen to make the symbol time on
each substream much greater than the delay spread of the
channel
Ø Make the substream bandwidth less then the channel
coherence bandwidth 7
Basic Multicarrier Transmitter
Ø a high rate stream of R bps is broken into L parallel streams each with
rate R/L and then multiplied by a different carrier frequency

8
A Basic Multicarrier Receiver
Ø each subcarrier is decoded separately, requiring L independent receivers

9
Channel Effects
Ø flat fading on each subchannel since B/L ≪ Bc, even though the overall
channel experiences frequency selective fading, i.e. B > Bc.

10
Possible but not practical
Ø a large bandwidth penalty will be inflicted since the
subcarriers can’t have perfectly rectangular pulse shapes
and still be time-limited
Ø very high quality (and hence, expensive) low pass filters
will be required to maintain the orthogonality of the the
subcarriers at the receiver.
Ø this scheme requires L independent RF units and
demodulation paths
11
OFDM Basics
Ø OFDM utilizes an efficient computational technique
known as the Discrete Fourier Transform (DFT), more
commonly known as the Fast Fourier Transform (FFT)
Ø No need for multiple radios

12
OFDM Basics
Ø L data symbols are grouped into a block – OFDM symbol
Ø Ts = symbol time for each data symbol
Ø T = LTs = OFDM symbol duration
Ø ! = Delay spread of the channel
Ø If guard time Tg > !, no interference between subsequent
OFDM symbols
Ø No ISI

13
Exploiting Properties of DFT
Ø Circular Convolution
Ø Frequency domain output
Ø It is ISI-free channel in the frequency domain, where each
input symbol X[m] is simply scaled by a complex-value
H[m]
Ø Note that the duality between circular convolution in the
time domain and simple multiplication in the frequency
domain is a property unique to the DFT.
14
Exploiting Properties of DFT
Ø L point DFT

Ø Inverse DFT (IDFT)

Ø At receiver, if channel frequency response H[m] is known,

input is derived as

15
How to create circular convolution in channel?
Ø Cyclic Prefix

Ø If max channel delay spread = !"#$ = v samples

Ø Then add a guard of v samples
Ø Time domain representation:

16
CP creates Circular Convolution
Ø Output of Channel: !"# = ℎ ∗ '"#

Ø for L ≫ v, the inefficiency due to the cyclic prefix can be

made arbitrarily small by increasing the number of
subcarriers
17
CP creates a circular convolution
Ø creates a circular convolution
at the receiver (signal y) even
though the actual channel
causes a linear convolution.

18
Penalties of CP
Ø v redundant symbols are sent
"#$
Ø Required bandwidth increases from ! to !
"
"#$
Ø Transmit power penalty 10'()*+ ,!
"
"
Ø Rate loss = Power loss =
"#$

19
Zero Prefix
Ø Null Guard Band
Ø At the receiver, the “tail” can be added back in
Ø Recreates the effect of a CP

()*
Ø Reduces Tx power by 10#$%&' +,
(
20
Zero Prefix Issues
()*
Ø Increases the receiver power by 10#$%&' +,
(
Ø With CP transmitted, the tail can be ignored
Ø Additional noise from the received tail symbols is added
back into the signal
Ø Higher noise power compared to transmitted CP
. ()* .
Ø - → -
(

21
Frequency Equalization
Ø data symbols are estimated using a one-tap frequency
domain equalizer

Ø !" is the complex response of the channel at the

frequency #$ + & − 1 ∆#

22
An OFDM System

23
An OFDM Transmitter

24
OFDM Parameters in LTE for 10MHz Channel

25
Timing Offset
Ø Cyclic Prefix provides some toleration in error in timing
synchronization

0 samples -8 samples

26
Frequency Offset
Ø carrier frequency/phase of transmitter’s local oscillator
(LO) and receiver’s LO can be off
Ø resulting frequency difference ΔFc Hz between
transmitter’s and receiver’s carrier introduces the
additional term ! "#$∆&'/&)* in the baseband multiplex
→ ICI (Inter-Carrier Interference)
Ø receiver needs to compensate this offset

27
ICI due to frequency offset
Ø Coarse correction
§ Short preamble based
Ø Fine correction
§ Long preamble based,
pilot tracking

28
Channel Fading and Recovery
Ø Recall: at receiver, if channel frequency response H[m] is
known, input is derived as
Ø OFDM is wideband
Ø Each SC is narrowband
Ø Flat fading on each SC
Ø But overall channel experiences
frequency selective fading
29
Fading across subcarriers
Ø transmit power on subcarrier ! is "#
Ø fading on that subcarrier is $#
Ø received SNR in subcarrier ! is %# = "# $#' /(*+ ,)
Ø where *+ is the noise power and , is the bandwidth

Ø Received SNR depends on $#

Ø $# varies with time in wireless channels
30
Frequency Equalization
Ø The fading !" is inverted in the receiver
Ø Received signal is multiplied by 1/!"
%& '& (
Ø Received signal power = *"
'& (
Ø Pros: removes the impact of fading
Ø Cons: it enhances the noise (incoming noise gets
multiplied by 1/!" )

31
Ideal Channel Estimation
Ø Wireless channels change frequently ~ 10 ms
Ø Require frequent channel estimation
Ø Many systems use pilot tones – known symbols
§ Given sk, for k = k1, k2, k3, … solve xk = ål=0L hl e-j2p k l/N sk for hl
§ Find Hk = ål=0L hl e-j2p k l / N (significant computation)
Ø More pilot tones
§ Better noise resilience
§ Lower throughput Pilot tones
magnitude

frequency
32
Channel Estimation Via Interpolation
Ø More efficient approach is interpolation
Ø Algorithm
§ For each pilot ki find Hki = xki / ski
§ Interpolate unknown values using interpolation filter
§ Hm = am,1 Hk1 + am,2 Hk2 + …
Ø Comments
§ Longer interpolation filter: more computation, timing sensitivity
§ Typical 1dB loss in performance in practical implementation
magnitude

frequency
33
Precoding or Pre-equalization
Ø Opposite of frequency equalization
Ø If the transmitter have knowledge of the subchannel fading !"
Ø Transmitter transmits #-th subcarrier signal with power
$" /!" &
Ø Channel gain !"
'( )( *
Ø Received signal power )( *
= $"
Ø Noise power is not multiplied
34
Adaptive Loading
Ø vary the data rate and power assigned to each subchannel
relative to that subchannel gain
Ø Variable rate variable power can be assigned to
subchannel to receive maximum capacity given a power
budget.

35
Peak to Average Power Ratio (PAPR)
Ø In time domain, OFDM is a sum of multiple narrowband signals.

36
High PAPR: Implementation challenges of OFDM
,-./0
Ø !"!# = 10'()*+
,/12
Ø generates out-of-band energy
(spectral regrowth) Typical Power Amplifier Response

Ø in-band distortion (constellation

tilting and scattering)

37
PAPR Reduction Techniques
Ø clipping and filtering
Ø selected mapping
Ø coding techniques

38
FFT Implementation
Ø The key to making OFDM realizable in practice is the utilization of the FFT algorithm
for computing the DFT and the IFFT algorithm for computing the IDFT, which reduces
the number of required multiplications and additions from O(L2) to O(L log L), which is
W 0·0 W 0·0 W 0·0

extremely significant. x(0)

16 8 4
⇥ ⇥ ⇥ X(0)
W 1·0 W 1·0 W 1·0
16 8 4
x(1) ⇥ ⇥ ⇥ X(8)
W 2·0 W 2·0 W 0·1
16 8 4
x(2) ⇥ 3·0 ⇥ 3·0 ⇥ 1·1 X(4)
W W W
16 8 4
x(3) ⇥ 4·0 ⇥ 0·1 ⇥ 0·0 X(12)
W W W
16 8 4
x(4) ⇥ 5·0 ⇥ 1·1 ⇥ 1·0 X(2)
W W W
16 8 4
x(5) ⇥ 6·0 ⇥ 2·1 ⇥ 0·1 X(10)
W W W
16 8 4
x(6) ⇥ 7·0 ⇥ 3·1 ⇥ 1·1 X(6)
W W W
16 8 4
x(7) ⇥ 0·1 ⇥ 0·0 ⇥ 0·0 X(14)
W W W
16 8 4
x(8) ⇥ 1·1 ⇥ 1·0 ⇥ 1·0 X(1)
W W W
16 8 4
x(9) ⇥ 2·1 ⇥ 2·0 ⇥ 0·1 X(9)
W W W
16 8 4
x(10) ⇥ 3·1 ⇥ 3·0 ⇥ 1·1 X(5)
W W W
16 8 4
x(11) ⇥ 4·1 ⇥ 0·1 ⇥ 0·0 X(13)
W W W
16 8 4
x(12) ⇥ 5·1 ⇥ 1·1 ⇥ 1·0 X(3)
W W W
16 8 4
x(13) ⇥ 6·1 ⇥ 2·1 ⇥ 0·1 X(11)
W W W
16 8 4
x(14) ⇥ 7·1 ⇥ 3·1 ⇥ 1·1 X(7)
W W W
16 8 4
x(15) ⇥ ⇥ ⇥ X(15)
39
Brief History of OFDM
Ø Although OFDM has become widely used only recently, the
concept dates back some 40 years.
§ 1958: The “Kineplex” system was developed, which was a multicarrier modem
for the HF bands (3 to 30MHz). This is widely considered the first ever
multicarrier system—it actually used multiple HF radios as the FFT was not re-
discovered9 until 1954.
§ 1966: Chang shows in the Bell Labs technical journal that multicarrier
modulation can solve the multipath problem without reducing data rate. This is
generally considered the first theoretical publication on multicarrier
modulation, although there were naturally precursory studies, including
Holsinger’s 1964 MIT dissertation and some of Gallager’s work on waterfilling.
40
Brief History of OFDM
§ 1971: Weinstein and Ebert show that multicarrier modulation can be
accomplished using a “Discrete Fourier Transform” (DFT).
§ 1985: Cimini at Bell Labs identifies many of the key issues in OFDM
transmission and does a proof of concept design.
§ 1993: DSL adopts OFDM, also called “Discrete Multitone,” following successful
field trials/competitions at Bellcore vs. equalizer-based systems.
§ 1999: IEEE 802.11 committee on wireless LANs releases 802.11a standard for
OFDM operation in 5GHz UNI band.
§ 2002: IEEE 802.16 committee releases OFDM-based standard for wireless
broadband access for metropolitan area networks under revision 802.16a.

41
Brief History of OFDM
§ 2003: IEEE 802.11 committee releases 802.11g standard for operation in the
2.4GHz band.
§ 2003: The “multiband OFDM” standard for ultrawideband is developed,
showing OFDM’s usefulness in low-SNR systems.
§ 2005: 802.16e standard is ratified, supporting mobile OFDMA for WiMAX.
§ 2006: First commercial LTE demonstrations by Siemens (now Nokia Siemens
Networks).
§ 2008: Qualcomm, the primary backer of Ultramobile Broadband (UMB), the
main future competition to LTE and WiMAX and also OFDM/OFDMA-based,
announces it will end UMB development and transition to LTE, solidifying LTE
as the leading beyond 3G cellular standard.
42
 Brief History of OFDM
§ 2009: 3GPP Release 8 LTE/SAE specifications completed and released.
§ 2009: 802.11n standard is ratified, which performs MIMO-OFDM for wireless
LANs for peak data rates of 600 Mbps.

S. B. Weinstein, "The history of orthogonal frequency-division multiplexing [History of Communications]," in IEEE Communications Magazine, vol. 47, no. 11, November 2009.
43
OFDMA
Ø Multiuser communication using OFDM in downlink LTE/5G

44
OFDMA
Ø Resource (OFDM subcarriers) can be allocated based on
the application, data rate and QoS requirements
Ø Allocate subcarriers based on user channel fading
§ Requires user feedback
Ø Subcarriers are modulated at different rates based on
received SNR at each UE

45
OFDMA Tx and Rx for Downlink

46
OFDMA unsuitable for uplink
Ø Uplink is naturally asynchronous - inevitable
time/frequency offsets from different UEs that transmit
simultaneously
Ø OFDMA: PAPR is a significant issue
Ø SC-FDMA (Single-Carrier Frequency Division Multiple
Access) is used for uplink
Ø Often called as DFT-coded OFDM
Ø Significantly lower PAPR than OFDMA
47
 SC-FDMA
In SC-FDMA, frequency domain
equalization is applied to each user’s
signal independently after the FFT

48
Modes of SC-OFDMA
Ø Interleaved SC-FDMA (IFDMA)
§ Subcarriers are equidistantly distributed
Ø Localized SC-FDMA (LFDMA)
§ Set of adjacent subcarriers

Ø IFDMA is less prone to transmission errors, channel

dependent scheduling of subcarriers can be done. 49
Time/Frequency Representation of SC-FDMA

50
Wireless Channel
Ø In discrete time, it is represented as a tap delay line
ℎ ", $ = ℎ& ' ", $ + ℎ) ' " − 1, $ + ⋯ + ℎ- ' " − ., $
Ø (. + 1) channel taps
Ø Channel is sampled at 12 = 1/4, 4 is symbol duration
Ø If channel is static over . + 1 4 seconds, output is
y ", $ = ∑:
789: ℎ ;, $ <[" − ;] = ℎ[", $] ∗ <["], ∗ denotes convolution

Ø In vector form, channel can be represented as a time-varying

. + 1 ×1 column vector
B C = ℎ& $ ℎ) $ … ℎ- $ E

51
Key Attributes of Channel
Ø What is the value for the total received power? In other words,
what are the relative values of the hi terms?
§ A number of different effects cause the received power to vary over long (path loss),
medium (shadowing), and short (fading) distances.
Ø How quickly does the channel change with the parameter t?
§ The channel coherence time specifies the period of time over which the channel’s
value is correlated. The coherence time depends on how fast the transmitter and
receiver are moving relative to each other.
Ø What is the approximate value of the channel duration ν?
§ This value is known as the delay spread, and is measured or approximated based on
the propagation distance and environment. 52
Free Space Path Loss
Ø Free space loss, ideal isotropic antenna
energy received at an antenna distance
( 4π d ) ( 4π fd )
2 2
Pt d away is inversely proportional to the
= =
Pr λ 2
c2 sphere surface area
!
o Pt = signal power at transmitting antenna
o Pr = signal power at receiving antenna
o λ = carrier wavelength
o d = propagation distance between antennas
o c = speed of light (3 ×108 m/s)
where d and λ are in the same units (e.g., meters)
Ø With antenna gains

53
Free Space Loss
Ø Free space loss equation can be recast:
Pt ⎛ 4π d ⎞
180

LdB = 10log = 20log ⎜ 170 Hz

⎝ λ ⎟⎠
00 G
f=3
Pr
!
160

150 z

= -20 log(l ) + 20 log(d ) + 21.98 dB

0 GH
f=3
140

130
GHz
f=3

Loss (dB)
æ 4pfd ö
÷ = 20 log( f ) + 20 log(d ) - 147.56 dB
120

= 20 logç 110
00 M
Hz

è c ø 100
f=3

90 Hz
0M
f=3
80

70

60
1 5 10 50 100
Distance (km)

54
Two Ray Ground Reflection

Ø Empirical Pathloss Formula

! = #$%ℎ'()) *+,(-*-%
./ = 11
#/ = 2*3*45*. ,(6*7 $% ./

55
Path Loss Exponent in practical systems

Table 6.5 Path Loss Exponents for Different Environments [RAPP02]

Environm ent Path Loss Exponent, n

Free space 2
Urban area cellular radio 2.7 to 3.5
Shadowed cellular radio 3 to 5
In building line-of-sight 1.6 to 1.8
Obstructed in building 4 to 6
Obstructed in factories 2 to 3

56
Shadowing
Ø Trees and buildings may be located between the
transmitter and the receiver and cause degradation in
received signal strength
Ø Shadowing is a random process

57
Fading

Ø Multipath
Ø Local Scattering
Ø Constructive &
Destructive Interference
58
Channel Impulse Response
Ø The channel is time varying, so the channel impulse response is also a
function of time and can be quite different at time t + Δt than it was at
time t

Channel Channel Impulse Response

59
Doppler Spread
Ø Doppler power spectrum is caused by motion between the transmitter
and receiver
Ø Doppler power spectrum gives the statistical power distribution of the
channel versus frequency for a signal transmitted at one exact frequency
Ø Doppler spread is where v is the maximum speed between the transmitter
and the receiver, fC is the carrier frequency, and c is the
speed of light

Ø Doppler varies with fc. If communication bandwidth B << fc, fD can be

treated as approximately constant.
Ø The coherence time and Doppler spread are also inversely related
60
3GPP Channel Model

61
Channel Parameters

62
RMS Delay Spread

63
Categories of Multiple Antenna Tx & Rx
Ø Spatial Diversity
§ a number of different versions of the signal to be Tx/Rx
§ provides resilience against fading
Ø Interference suppression
§ uses the spatial dimensions to reject interference from other users
§ through the physical antenna gain pattern or through other forms of array processing
such as linear precoding, postcoding, or interference cancellation
Ø Spatial multiplexing
§ allows multiple independent streams of data to be sent simultaneously in the same
bandwidth, and hence is useful primarily for increasing the data rate
64
Spatial Diversity – Array Gain
Ø Coherently combines energy of each antenna (channels can be correlated
if LOS and closely spaced antenna)
Ø Noise is uncorrelated and do not add coherently
Ø In correlated flat fading channel, received SNR increases linearly with
the number of receive antennas, Nr

!" = ℎ" % + '" = ℎ% + '" , ℎ is correlated flat fading channel

SNR at antenna i is ;" = ℎ< /> <
DE DE
Resulting Signal from all antennas ! = ∑"BC !" = FG ℎ% + ∑"BC '"
DE L M DE LM
Combined SNR is γ = KDE NM = = FG ;"
NM 65
Spatial Diversity – Diversity Gain
Ø Channel varies over space
Ø rms angular spread of a channel = !"#$ = statistical distribution of the
angle of the arriving energy
Ø Dual of angular spread is coherence distance, Dc
Ø A coherence distance of d means that any physical positions separated
by d have an essentially uncorrelated received signal amplitude and
phase
Ø %& ≈ .2*/!"#$ , in Rayleigh fading, %& ≈ 9*/16/
Ø coherence distance increases with the carrier wavelength λ, so higher-
frequency systems have shorter coherence distances
66
Spatial Diversity – Diversity Gain
Ø If Nt transmit antennas and Nr receive antennas that are
sufficiently spaced are added to the system
Ø the diversity order is Nd = NrNt
Ø Nd is the number of uncorrelated channel paths between the
transmitter and receiver
Ø probability of all the Nd uncorrelated channels having low SNR
is very small
Ø bit error probability improves dramatically
67
 Spatial Diversity – Diversity Gain

Sufficient spacing for the antennas is critical

for increasing the system reliability 68
Benefits of Spatial Diversity
Ø Increased data rate
§ Antenna diversity increases SNR linearly
§ Receiver techniques increase capacity logarithmically wrt #antennas
§ data rate benefit rapidly diminishes as antennas are added
§ Multiple independent streams increase aggregate data rate
Ø Increased coverage or reduced Tx power
§ With only array gain, increase in SNR is !" #$
§ Increase in SNR increases coverage range
§ transmit power can be reduced by 10'()*+!" ,-
69
Receive Diversity
Ø Receive multiple streams and combine them
§ Selection Combining
§ Maximal Ratio Combining
§ Equal Gain Combining
§ Hybrid Combining

70
Selection Combining
Ø estimates the instantaneous strengths
of each of the Nr streams and selects
the highest one
Ø Since it ignores the useful energy on
the other streams, SC is suboptimal
Ø Its simplicity and reduced hardware
requirements make it attractive in many
cases

71
Maximal Ratio Combining
Ø use linear coherent combining of
branch signals so that the output
SNR is maximized
Ø Individual branch signal:

Ø Output of the combiner:

Ø Best performance
Ø coherent technique, i.e., signal’s Ø Lot of circuitry for individual
phase has to be estimated receivers 72
Equal Gain Combining
Ø corrects only the phase
Ø Simpler than MRC, easier to implement

Ø Hybrid Combining
§ Combination of multiple of combining techniques

73
Comparing Receiver Diversity

74
Transmit Diversity
Ø signals sent from different transmit antennas interfere with one another
Ø processing is required at both the transmitter and the receiver
Ø goal is to achieve diversity while removing or attenuating the spatial
interference
Ø used for the downlink of infrastructure-based systems
Ø Mobile stations may not need to use it due to size, power constraints
Ø Can be open loop or closed loop

75
Open Loop Transmit Diversity
Ø Space Time Block Codes (STBC)
Ø Alamouti code is a type of STBC
Ø ease of implementation—linear
at both the transmitter and the
receiver

76
Alamouti Code
Ø If two symbols to be transmitted

Ø Received Signal, (flat fading channel & h1 (t = 0) = h1 (t = T ) = h1 )

Ø Linear diversity Combining (channel known to receiver)

Ø Eliminates spatial interference 77

STBC in OFDM
Ø Owing to the flat-fading assumption, the STBC in an
OFDM system is performed in the frequency domain,
where each subcarrier experiences flat fading
Ø Space/time trellis codes introduce memory and achieve
better performance (about 2dB) than orthogonal STBCs
Ø Trellis code decoding complexity ! "#$% &',&)
Ø STBC complexity !(+,- ./ , .0 )

78
Alamouti STBC vs MRC
Ø Alamouti STBC outperforms
MRC at high SNR owing to
the diversity order
Ø MRC has better BEP
performance than Alamouti
STBC at low SNR owing to
the array gain

79
Closed loop Transmit Diversity
Ø Feedback needs to be added to the system
Ø channel changes quickly in a highly mobile scenario
Ø closed-loop transmission schemes feasible primarily in
fixed or low-mobility scenarios

80
Transmit Selection Diversity
Ø A subset of all available antennas used
Ø Subset corresponds to the best channels between the
transmitter and the receiver
Ø Advantages:
§ significantly reduced hardware cost and complexity
§ reduced spatial interference, since fewer transmit signals are sent
§ reaches Nt Nr diversity order, even though only a subset of all antennas
are used
81
Linear Diversity Precoding
Ø general technique for improving the data rate by
exploiting the CSI at the transmitter
Ø diversity precoding, a special case of linear precoding,
where data rate is unchanged
Ø linear precoder at the transmitter and a linear postcoder
at the receiver

82
Received Data Vector
Ø ! = #$ = #(&'( + *)
§ M is the number of spatial data “streams” sent
§ Transmitted vector ( is ,×1
§ Received vector $ is /0 ×1
§ Postcoder matrix # is ,×/0
§ Channel matrix & is /0 ×/1
§ Precoder matrix F is /1 ×,
§ M = 1 is known as maximal ratio transmission (MRT)

83
Precoding in MIMO OFDM

84
Interference Cancellation Suppression
Ø Suppress undesired signals and/or enhance the power of the desired
signal
Ø In MIMO, channel is multidimensional
§ the dimensions of the channel can be applied to null interference in a certain direction, while
amplifying signals in another direction
§ Contrast to transmit diversity (statistical diversity of the total signal is increased)
Ø Types:
§ DOA-Based Beamsteering
§ Linear Interference Suppression: Complete Knowledge of Interference Channels

85
Beamsteering (Physically steering)
Ø Electromagnetic waves can be physically steered to create
beam patterns at either the transmitter or the receiver
Ø Static pattern-gain beamsteering : called sectoring
§ Example: in a three-sector cell, a strong beam is projected over 120
degrees, while very little energy is projected over the remaining 240
degrees

86
DOA based Beamsteering
Ø Incoming signal may consist of
§ desired energy + interference energy (other users or multipath)
Ø Signal processing techniques are used to identify angle of
arrival (AoA) of these signals
§ MUSIC, ESPIRIT, JADE, MLE
Ø These AoAs are used by a beamformer to calculate
weighting vector of the antenna elements

87
Uniform Linear Array
Ø wave at the first antenna element travels an
additional distance of d sin θ to arrive at the second
element
Ø difference in propagation distance between the
adjacent antenna elements results in arrival-time
delay, τ = d/c sin θ
Ø signal arriving at the second antenna can be
expressed in terms of signal at the first antenna
element

88
Uniform Linear Array
Ø For an antenna array with Nr elements all spaced by d ,
the resulting received signal vector is

a(θ) is the array response vector

89
Weight vector Calculation
Ø Example:
§ a three-element ULA with d = λ/2
§ desired signal is received at θ1 = 0, two interfering signals at θ2 = π/3
and θ3 = –π/6

Ø Objective:
§ The beamforming weight vector w = [w1 w2 w3 ]T should increase the
antenna gain in the direction of the desired user while minimizing the
gain in the directions of interferers.
90
Weight vector Calculation
Ø weight vector w should satisfy the following criterion

Ø Solution for weight vector

91
Null-steering Beamformer
Ø number of nulls is less than the number of antenna elements.
Ø the antenna gain is not maximized at the direction of the
desired user
Ø trade-off between interference nulled and desired gain lost
Ø May exist several unresolved components coming from
significantly different angles
Ø DOA-based beamformer is viable only in
§ LOS environments or
§ in environments with limited local scattering around the transmitter
92
Linear Interference Suppression
Ø Received signal vector

Ø where
§ wt is the Ntx1 weighting vector at the desired user’s transmitter,
§ x is the desired symbol
§ xI = [x1 x2 … xL]T is the interference vector
§ n is the noise vector
§ H is the Nr x Nt channel gain matrix for the desired user
§ HI is the Nr x L channel gain matrix for the interferers
93
Linear Interference Suppression
Ø With statistical knowledge of channel:
§ In order to maximize the output SINR at the receiver, joint optimal weighting vectors
at both the transmitter and the receiver can be obtained
Ø This is termed optimum eigenbeamformer, or interference-
aware beamforming, or optimum combiner (OC)
Ø interference-aware beamformer is conceptually similar to the
linear diversity precoding
Ø difference is that the eigen-beamformer takes interfering
signals into account
94
Spatial Multiplexing

Ø Nt <= Nr
Ø Split the incoming high rate-data stream into Nt independent data
streams
Ø decoding Nt streams is theoretically possible when there exist
at least Nt nonzero eigenvalues in the channel matrix, that is
rank(H) ≥ Nt
Ø Assuming that the streams can be successfully decoded, the nominal
spectral efficiency is thus increased by a factor of Nt 95
Spatial Multiplexing: Key Points
Ø When the SNR is high, spatial multiplexing is optimal.
§ The capacity, or maximum data rate, grows as min(Nt, Nr) log(1 + SNR)
when the SNR is large.
Ø When the SNR is low, the capacity-maximizing strategy is
to send a single stream of data using diversity precoding.
§ Although the capacity is much smaller than at high SNR, it still grows
approximately linearly with min(Nt, Nr) since capacity is linear with SNR
in the low-SNR regime.

96
Spatial Multiplexing: Key Points
Ø Both of these cases are superior in terms of capacity to space-
time coding, where the data rate grows at best logarithmically
with Nr
Ø The average SNR of all Nt streams can be maintained without
increasing the total transmit power relative to a SISO system
§ each transmitted stream is received at Nr ≥ Nt antennas and hence recovers the
transmit power penalty of Nt due to the array gain.
Ø Note: even a single low eigenvalue in the channel matrix can
dominate the error performance.
97
Open Loop Spatial Multiplexing
Ø Optimal Receiver:
§ Maximum likelihood: finds input symbol most likely to have resulted in
received vector
§ Exponentially complex with # of streams and constellation size
Ø Sphere Decoder:
§ Only considers possibilities within a sphere of received symbol.
o If minimum distance symbol is within sphere, optimal, otherwise null is returned

xˆ = arg min | y - Hx |2 xˆ = arg min | Q H y - Rx |2

x x:|Q H y - Rx|< r
98
Linear Detectors : Zero Forcing Detector
Ø sets the receiver equal to the
inverse of the channel Gzf = H–1

Ø zero-forcing detector removes the spatial interference from

the transmitted signal
Ø As Gzf inverts eigenvalues of H, poor subchannels can severely
amplify noise
Ø Not practical in interference-limited MIMO (LTE)
99
Linear Detectors : MMSE Receiver
Ø MMSE receiver attempts to strike a balance between spatial-interference
suppression and noise enhancement by minimizing the distortion

Ø As the SNR grows large, the MMSE detector converges to the ZF detector
Ø At low SNR, it prevents the worst eigenvalues from being inverted

100
Interference Cancellation: BLAST
Ø Bell labs LAyered Space-Time (BLAST) : invented and prototyped in Bell
Labs
Ø BLAST consists of parallel “layers” supporting multiple simultaneous
data streams
Ø The layers (substreams) in BLAST are separated by interference-
cancellation techniques that decouple the overlapping data streams
Ø two most important techniques are
§ the original diagonal BLAST (D-BLAST)
§ its subsequent version, vertical BLAST (V-BLAST)

101
D-BLAST
Ø in each layer’s data is transmitted in a diagonal of space and
time
§ groups the symbols into “layers” that are then coded in time independently of the
other layers
§ these layers are then cycled to the various transmit antennas in a cyclical manner
Ø one layer decoded at a time
Ø Each successive layer is detected by
§ nulling the layers that have not yet been detected
§ canceling the layers that have already been detected
102
D-BLAST Pros & Cons
Ø Pro: each symbol stream achieves diversity
§ in time via coding and
§ in space by it rotating among all the antennas
Ø Cons:
§ Decoding process is iterative and complex
§ wastes space/time slots at the begin- ning and end of a D-BLAST block

103
V-BLAST
Ø each antenna transmits an independent symbol stream—for
example, QAM symbols
Ø different techniques can be used at the receiver to separate the
various symbol stream from one another
§ Including ZF, MMSE
§ the strongest symbol stream is detected, using a ZF or MMSE receiver
§ subtracted out from the composite received signal
Ø Pros:
§ ordered successive interference cancellation lowers the block error rate by a factor of ten relative
to a purely linear receiver

Ø Cons:
§ error propagation when initial layers are detected incorrectly leads to huge penalty
§ depends on high SNR (not available in cell edge)
104
 Closed Loop Spatial Multiplexing
Ø The advantage of channel knowledge
Ø SVD Precoding and Postcoding
§ Channel expressed as singular-value decomposition (SVD, or generalized eigenvalue
decomposition)
§ U and V are complex unitary matrices, is a diagonal matrix of singular values (non-
negative real numbers)
Impractical, but promising results
compared to open loop approach
complexity of finding the SVD of an
NtxNr matrix is on the order of O(Nr,Nt2)
if Nr >= Nt
https://en.wikipedia.org/wiki/Singular_value_decomposition
105
Linear Precoding and Postcoding
Ø ! = #$ = #(&'( + *)
Ø decomposes the MIMO channel into a set of
§ M is the number of spatial data “streams” sent
parallel subchannels § Transmitted vector ( is ,×1
Ø the precoder and the postcoder can be jointly § Received vector $ is /0 ×1
designed based on § Postcoder matrix # is ,×/0
§ Channel matrix & is /0 ×/1
§ information capacity, error probability, detection MSE, or received
SNR § Precoder matrix F is /1 ×,
§ 1 ≤ M ≤ min(/0 , /1 )
Ø precoder weights are used to maximize the total
capacity by distributing more transmission power to
subchannels with larger gains and less to the
others - waterfilling

106
How to choose MIMO Techniques?
Ø https://ieeexplore.ieee.org/document/5374062
Ø Due March 25 after Spring break

107
Channel Estimation for MIMO OFDM
Ø Channel estimation required
§ At the receiver in order to
o coherently detect the received signal
o for diversity combining
o spatial-interference suppression
§ At the transmitter
o For closed loop MIMO

Ø Types:
§ Training based – known symbols (preambles, pilots) transmitted, reliable, mostly used
§ Blind – no training, no overhead, low convergence speed, lower estimation accuracy
108
Training Symbols
Ø Two ways to transmit training symbol:
§ Preambles : send a certain number of training symbols prior to the user
data symbols
§ Pilot tones : insert a few known (time, frequency, phase, amplitude) pilot
symbols among the subcarriers
Ø Channel estimation typically done by using
§ the preamble for synchronization and initial channel estimation
§ the pilot tones for tracking the time-varying channel in order to
maintain accurate channel estimates
109
Pilot Insertion Patterns
Ø received signal at each antenna is a superposition of the signals transmitted from Nt
transmit antennas
Ø the training signals for each transmit antenna should not interfere with one another
Ø Independent: orthogonality achieved in time domain, requires Nt training signal times
Ø Scattered: orthogonality achieved in frequency domain
Ø Orthogonal: orthogonality achieved using orthogonal codes

110
Time Domain Channel Estimation
Ø Preamble based with cyclic prefix
x(l) is the lth time sample of the
transmitted OFDM symbol, and h(i) is the
ith time sample of the channel impulse
response

X is deterministic and hence known a priori by the receiver

111
Frequency Domain Channel Estimation
Ø simpler in the frequency domain than in the time domain

112
Equalization
Ø Linear Equalization
§ runs the received signal through a filter that models the inverse of the
channel
Ø Non-linear Equalization
§ uses previous symbol decisions made by the receiver to cancel out their
subsequent interference and so are often called decision-feedback
equalizers (DFEs)
Ø Maximum-likelihood sequence detection (MLSD)
113