Professional Documents
Culture Documents
2
Small antennas - Introduction
• In microstrip
technology
• Antenna pattern
3
Small Antennas for mobile phones
• Balanis
4
Antennas for macrocells
• Spatial
diversity
• CDF
• Gain
6
Smart spectrum analyzer up to 10
GHz
- C programmable ultra
portable spectrum
analyzer
- USB data exchange
- AAronia (Germany)
7
Ultra portable and programmable
Vector Network Analyzer kit
Specifications:
• Frequency range 1MHz – 3GHz
• Calibration using open-short-load for
accurate results
• Range of Z from 1 to 1000 ohm
• Two ports VNA with S11 and
S21; displayed and save results
• RF Generator with output power of -6dBm
@ 500 MHz
• Powered from USB
• SMA connectors for better isolation
• Dynamic range – up to 70 dB @ 500 MHz
• Boot loader for future firmware upgrades
• User friendly interface for PC Windows,
Linux and Mac
• Integrated Smith chart in software
• Android Mobile Phone software
• Export data in several formats JPEG, 8
EXCEL, ZPLOT, S2P, PDF
LEAT- ASL Antennas
- Microstrip antenna for:
- 3G & 4G systems UMTS & LTE
- Wireless Network sensors
- Array antennas measurements for
LTE systems simulation and arrays
processing in MATLAB
9
Texas instruments embedded kit
- 868-915 MHz
- 433-442 MHz
- 2.4-2.5 GHz
- C programmable
- Antenna testing
11
RFID - Sensors
12
RFID – For deposit management
13
ASL - RFIDs
14
GPS – Sensors on RASPBERRY PI 3-B
15
Software defined radio - SDR
16
Open source GNU Radio under Linux – SDR –
Concatenated coding and OFDM for WiMAX
17
Examples: LTE Antenna beamforming effect over the
64QAM constellation
% Set PDSCH beamforming
vector
if (ntx==1)
% Use equal
transmission gains for each
antenna
rmc.PDSCH.W = [17 17
17 17]/34;
else
% Beamforming vector
matched to channel response
rmc.PDSCH.W = [17 8-
15*1i -8+15*1i 15+8*1i]/34;
end
18
Powerful FEM simulator
- HFSS v13
- Small scale simulation
- Antenna design in
microwave
- Simulations
- Optimization
- Antenna
parameterization
- Implementation
- Laboratory works
- Semester thesis
- Dissertation
19
Powerful FEM/SYS simulator
- ADS v2013.06
- Systems
- Simulations
- Array network design
- Implementation
- 15 licenses for ASL use
only
20
FEKO – Momentum/FEM
• Large scale antenna
modeling
• Automotive
• Radar
21
Compact yet powerful FEM/SYS
simulator
- SystemVue v2013.01
- Systems simulations
- Array network design
- Implementation
- Semester Thesis
- Dissertations
22
Academic SYS simulator
- Matlab vR2014a
- LTE Systems
- Simulations
- Array network design
- Toolboxes:
- LTE system toolbox
- Phased Array System
toolbox
- Semester Thesis
- Dissertations
23
Multiband
GSM / 3G / LTE
antenna design example
Targeted Parameters:
• S11 <-6dB in each band
• Antenna radiation efficiency >80%
Notes:
•The feeding and ground pins have arbitrary positions
•To avoid a large capacitance between the two GND planes on the main PCB many viases can be drawn arbitrary
•The antenna material can be copper
•Antenna bending is allowed. Freedom to choose its size.
•Parasitic elements can be applied on one face or the other of the vertical PCB to extend bandwidth
•Minimum antenna copper widths >0.5mm
•Use parameters for antenna elements size ->Optimetrics can be used to fine-tune the antenna
GSM / 3G antenna 824 – 960 MHz &1710 – 2170 MHz
Problem 1:
Design a dual band IFA antenna on the vertical PCB from the below picture
PCB – FR4 (thickness: 1,6mm; εr= 4.4; loss tangent= 0.02; 2 ground layers top and bottom on main PCB)
60mm
20mm 125mm
85mm
Antenna design for automotive challenge – Continental Timisoara
- Write scientific
papers in journals
(easy!)
27
Antenna coordinates
28
• Antenna feed terminal
29
EM wave
30
E field along the transmission line
31
- Front wave is perpendicular to the direction of propagation
60Pt
E 2 Erms
R
32
33
Pattern along the three regions
34
Antenna basics
35
Antenna parameters
1. Imput impedance
2. Bandwidth
3. Radiation pattern
4. Beamwidth
5. Imput impedance vs radiation
6. Polarization
7. Equivalent area, Gain and Directivity
8. SWAP – Size weight and power
36
Antenna parameters
1. Input impedance
Is a measure of the total opposition to the current flow and is evaluating
by two components (real + imaginary)
1
Z in Z 0 R jX
1
1
VSVR 1 Voltage standing wave ratio
1 RL[ dB ]
10 20
1
VSWR RL[ dB ]
RLdB 20 lg return loss 10 20
1
RL[ dB ]
Reflection coefficient
10 20
37
Impedance matching (Important!)
38
• Insertion loss
Insertion loss is defined as the ratio of power received at the end of the line and
transmitted power source line. It is a size that characterizes the line in terms of
losses incurred by the applied signal. In decibels losses are estimated to be as
small as possible (0 dB). Its determination taken into account final length of the
transmition line, so the effective value is expressed in terms of the line length in
dB / m and is positive (the argument of the logarithm lower than one).
𝑃
𝐼𝐿[𝑑𝐵] = −10𝑙𝑔 𝑃𝑅> 0
𝑇
39
• Return loss
Insertion loss is defined as the ratio of power received at the end of the line and
transmitted power source line. It is a size that characterizes the line in terms of
losses incurred by the applied signal. In decibels losses are estimated to be as
small as possible (0 dB). Its determination taken into account final length of the
transmition line, so the effective value is expressed in terms of the line length in
dB / m and is positive (the argument of the logarithm lower than one).
40
Relations between return loss, SWR and reflection coefficient
Convertion table: RL,
SWR and Г
41
2. Bandwidth
42
Calculation of the antenna frequency band by using the RL
if an antenna has two resonant frequencies (dual band)
43
3. Radiation pattern
An antenna radiation pattern or antenna pattern is defined as “a mathematical function
or a graphical representation of the radiation properties of the antenna as a function
of space coordinates. In most cases, the radiation pattern is determined in the far-field
region and is represented as a function of the directional coordinates.
Directivity
D ,
puterea radiata pe unitatea de unghi solid
puterea medie pe unitatea de unghi solid
dPrad d dP d
4 rad
Prad 4 Prad
d sin d d
44
Antenna pattern examples
45
46
Example – Pyramidal horn antenna simulated in HFSS
47
4. Beamwidth
-The beamwidth of a pattern is defined as the angular separation between two identical
points
on opposite sides of the pattern maximum (Pmax – 3 dB HPBW).
SLL-Side
lobes level
Pr Pd
Rr 2 2 Rin Rr 2 2
I0 I0
P Pd 2 j WH WE
2
36,565 I 0
Rin r 2 73,13
I 0 I 0* 2 I0
2
Pd 0
Rin Rr 2 Rr - For a dipole (unfolded)
I0
2
8 .r0
49
• Impedance along the dipol
50
6. Polarization
•The antenna is responsible for the direction and orientation of the vector E (electric field intensity)
since the wave is generated in the near field region. If the vector is acting in the horizontal plane, the wave
is horizontally polarized, and if it is in the vertical plane, polarization is vertical.
51
Polarization: elliptical, circular and linear
52
7. Equivalent area (aperture) and the gain
The power capturing characteristics of the antenna when a wave impinges on it. The
ratio of the available power at the terminals of a receiving antenna to the power flux
density of a plane wave incident on the.
PR R AeR
53
DT and DR are the antenna directivities of transmitter and receiver respectively and AeT
and AeR are equivalet areas of them.
PT DT
R 0 DT
4R 2
4 Aem
D
2
PT 4 4
G D rad D rad Aem Ae
PRF 2
2
54
8. SWAP – Size weight and power
55
Microstrip antenna technologies
1. Patch antenna
USAGE
It is one of the most common enspecialy at microwave frequencies where wireless
equipments work (mobile phones, WiFi networks, laptops and so on).
56
•Designing the patch
There are two methods that stands for the microstrip size calculation: lines model and
cavity model. The most used is the first which sees the antenna as an open transmition
line.
1 2
r 1 r 1 12h - Effective dielectric permitivity
eff 1
2 2 W
W2
G pentru W 0 - Load conductance
9002
W2
G pentru W 0
12002
58
eff 0,3 W h 0,264
L 0,412h - corections
eff 0,258 W h 0,8
g 0
L 2L - Line must be lengthen with something
2 2 eff
c - Resonance
fr
2 eff L L
1 / 2
c r 1 - Line width
W
2 fr 2
59
Exemple – Design a microstrip patch antenna at 1.6 GHz on a FR4 PCB with
electric permitivity εr = 4.4 having the thickness h = 1.524 mm. The fed line of the
antenna has 50 ohms characteristic impedance. After the antenna dimention
match the antenna to the 50 ohm characteristic impedance of the fed line.
c
0 18,69 cm
fr
0 r 1 1 / 2
W 5,7 cm
2 2
1 2
r 1 r 1 12h
eff 1 4,2
2 2 W
60
0
L 2L 4,43 cm
2 eff
4502
Rin 486 ohmi
2
W
61
• Feeding methods
Input impedance has a real value only at the resonance, this value depends by the conductance
value from its sides. Mostly the main problem emerges due to the mismatch between the feeding line
and the input impedance of the antenna. In this case a matching impedance must be consider using
stubs on microstrip PCB. Another technique is to resize the width W of the antenna to match the 50
ohms characteristic impedance of the feeding line.
The antenna admittance is of the form:
W2
Yin 2G
4502
4502
Z in Rin 50 Ω.
2
W
62
In order to obtain an input impedance of 50 ohms : W 0,95.... 1,20
W
h g ln g / h 1
A microtrip patch antenna usually has a reduced aperture even though its radiation efficiency is very high
(tends to the unity). When W increases the antenna gain will increase, as well. When we can not modify the W
parameter the inset feeding method is an option to keep the impedance matching. The inset feeding technique
exploit the variation of the impedance along the antenna (along z axes) variation that follows a co-sinusoidal
law :
z
Rin _ inset Rin cos2 Where z is the inset coordinate along the antenna length
L referring to one of its end
63
This program interface shows this variation of
the impedance along the length of the antenna
with z.
Rin _ inset
L 1
z cos
Rin
64
• Asymmetric feed line
method
If d is the smallest distance from the edge then the
input admittance could be equated as follows:
1
eff G 2 B 2 eff
Yin 2G cos2 2 d sin 2 4 d
Y02
0 0
Y = G+jB
2 eff
B L
g Z0
65
•Ansoft interface – Ansoft Designer is the own
by the Ansys Coorporation
66
Types of antenna feeding techniques a. coaxial fed b. EM fed
67
Electric field along the patch
Electric field bends to the antenna sides making it to looks longer from electric point of view. Voltage wave is
maximum at the ends and the current into the middle. Radiation tends to its maximum in the middle where the
current is maximum too.
68
•slot coupled antennas
A slot can be used every time the line couldn’t be attached phisically to the patch.
• VSWR controll
69
S11 vs frequency
70
From figure we can approximate antenna bandwidth at 18 MHz. The size of the slot can modify the WSVR
value.
71
S11 on Smith chart
72
3D directivity using wire-frame method Slot cupled
73
•Half patch antenna – PIFA (Planar Inverted F Antenna
- In the middle the input impedance is zero.
- Symmetry
- Cut the right side and use a short circuit into the middle to force a null
- We can shorten the antenna to the half (quarter wavelength)
74
- Half patch antenna in HFSS
75
•Left or right handed polarization
76
•SAR – Specific absorption rate
- Units [W/Kg].
- Calculation
𝜎 𝑟 𝐸 𝑟 2
SAR= 𝑑𝑟
𝜌 𝑟
-𝜌 𝑟 tissue density
- 𝐸 𝑟 2 electric field (RMS)
-σ 𝑟 electrical conductivity
77
- After 6 minutes of exposure
78
Estimate SAR with HFSS following the FCC – Federal
Communications Commission restrictions
79
Measurements and simulations comparisons of
SAR with HFSS
80
SAR – RMS
81
Arrays
- EM antenna array
-Smart antennas
82
-The very firs applications of
the EM arrays was in WWII
Modern applications
- avionics
83
- Equipment is at the cutting edge
of technological advanced
- In order to control the magnitude and phase for each radiator
the cutting edge of engineering is involving like DSP processing,
and software advanced algorithms. A sophisticated software
defined radio platform is the key.
84
Our technology
85
•A smart antenna is comprised by more arrays of identical radiators (small
antennas) which are connected by means of coaxial or microstrip lines. In the far
field region the resulting field emerges under the effect of the superposition effect
of each radiator.
•With the help of these lines the direction of the radiation is under the control of
the signal delay and gain (complex weights) along each path.
Conclusions
• The array could be linear (ULA-
Uniform linear array) most simple,
rectangular or circular (geometry)
• To be “smart”, there must be
some processing involved before
signal summation
86
•Antenna feeding network for ULA
87
• ULA-Uniform linear array
88
• Retele de antene de tip array linear
2𝜋𝑑 cos 𝜃0
𝛿 = 𝛽𝑑 cos 𝜃0 = - Equivalent phase difference
𝜆
between two consecutive
elements
89
• ULA – Superposition effect of the fields
𝐸𝑡𝑜𝑡𝑎𝑙
𝑒𝑥𝑝 −𝑗 𝛽 𝑟1 − 𝛿1 𝑒𝑥𝑝 −𝑗 𝛽 𝑟2 − 𝛿2
= 𝐸1 + 𝐸2
4𝜋 𝑟1 4𝜋 𝑟2
𝑒𝑥𝑝 −𝑗 𝛽 𝑟𝑁 − 𝛿𝑁
+ … + 𝐸𝑁
4𝜋 𝑟𝑁
E is the magnitude of the electric field
𝑁
𝑒𝑥𝑝 −𝑗𝛽 𝑟
𝐸𝑡𝑜𝑡𝑎𝑙 = 𝑓 𝜃, 𝜓 𝑒𝑥𝑝 −𝑗 𝑖 − 1 𝛽𝑑 cos 𝜃0 + 𝛿
4𝜋 𝑟
𝑖=1
Assumptions:
𝑒𝑥𝑝 −𝑗𝛽 𝑟
𝐸𝐹 = 𝑓 𝜃, 𝜓 element pattern or element factor
4𝜋 𝑟
𝑁−1
𝐴𝐹 = 𝑘=0 𝑒𝑥𝑝 −𝑗𝑘𝜑 = 𝑒 𝑗𝜑1 + 𝑒 𝑗𝜑2 +…+𝑒 𝑗𝜑𝑁
91
• ULA - Phase array
𝐸𝑡𝑜𝑡𝑎𝑙 = 𝐸𝐹 ∗ 𝐴𝐹
When phase difference is −𝛽𝑑 sin 𝜃 then the main lobe is beam
steered (is controlled) through 𝑛
92
•Linear array with two elements
d
𝑟1 ≈ 𝑟 + sin 𝜃 Distance
2 vectors
d
𝑟2 ≈ 𝑟 − sin 𝜃 magnitude
2
93
• Phase array
𝛿 𝛿
−𝑗 𝑗
-the phasor current in element 1 is 𝐼1 = 𝐼0 𝑒 2 and for the second is 𝐼2 = 𝐼0 𝑒 2
𝐼1 𝐼2
94
•Example – two elements array
d - is antenna spacing
The wave front is coming under theta angle from the right
𝐴𝐹 = 𝑒 𝑗𝜑1 + 𝑒 𝑗𝜑2
𝛽𝑑 𝜋 𝛽𝑑 𝜋
−𝑗 cos −𝜃 𝑗 cos −𝜃
𝐴𝐹0 = 𝑒 2 2 +𝑒 2 2
Wave front
𝑛
𝛽𝑑 𝛿 𝛽𝑑 𝛿
−𝑗 2 sin 𝜃−2 𝑗 2 sin 𝜃−2
𝑟2 AF = 𝑒 + 𝑒
𝑎2 𝜃 𝑎1
𝜃 𝜃0
𝐴𝐹0 𝛽𝑑 𝛿
d
𝛿 𝛿 sin 𝜃 − = 0
−
2 2 2 2
0 𝑟1
−𝑑 2 𝑑 2
AF 𝛽𝑑 sin 𝜃 = 𝛿
Trough receiver
95
•N elements uniform linear array
96
•N elements uniform linear array
𝑁 𝑁
𝐴𝐹 = 𝑒𝑥𝑝 𝑗 𝑖 − 1 𝛽𝑑 sin 𝜃 + 𝛿 = 𝑒𝑥𝑝 𝑗 𝑖 − 1 𝛼
𝑖=1 𝑖=1
𝑇
𝑎 θ = 1 𝑒𝑥𝑝 𝑗 𝛽𝑑 sin 𝜃 + 𝛿 … . . 𝑒𝑥𝑝 𝑗 𝑁 − 1 𝛽𝑑 sin 𝜃 + 𝛿
97
•N elements uniform linear array
- Geometric sequence
98
•N elements uniform linear array
𝑁 𝑁 𝑁
𝑒𝑥𝑝 𝑗 𝛼 𝑒𝑥𝑝 𝑗 𝛼 − 𝑒𝑥𝑝 −𝑗 𝛼
𝑒𝑥𝑝 𝑗𝑁𝛼 − 1 2 2 2
𝐴𝐹 = =
𝑒𝑥𝑝 𝑗𝛼 − 1 𝛼 𝛼 𝛼
𝑒𝑥𝑝 𝑗 2 𝑒𝑥𝑝 𝑗 2 − 𝑒𝑥𝑝 −𝑗 2
𝛼 𝛼 𝛼
Knowing that: 𝑒𝑥𝑝 𝑗 = cos + 𝑗 sin
2 2 2
𝑁
𝑁−1 sin 𝛼 - The term encircled with red is
𝐴𝐹 = 𝑒𝑥𝑝 𝑗 𝛼 2
𝛼 1, whenever the reference is in
2 𝑠𝑖𝑛 2 the middle of the array
99
•N elements uniform linear array
𝛼
sin 𝑁 2
𝐴𝐹 = 𝛼 - The maximum value is N
𝑠𝑖𝑛 2
𝛼 𝛼
𝑠𝑖𝑛 ≈ When alpha is small
2 2
𝛼
𝐴𝐹 sin 𝑁 - Normalized array factor
𝐴𝐹𝑛 = = 2
𝑁 𝛼 - Maximum is 1
𝑁2
100
- Operation modes
𝑑 𝑑
-AF for four elements array 𝜆
= 1 (left) and 𝜆
= 0,75 (right) 101
- Operation modes
𝑑 𝑑
-AF for four elements array 𝜆
= 0.5 (left) and 𝜆
= 0.25 (right) 102
END-FIRE mode - The name end-fire indicates that this
array’s maximum radiation is along the axis containing
the array elements. Thus, maximum radiation is “out the
end” of the array.
103
- Four elements array in the same condition as previous
ones
It should be noted that the mainlobe width for the ordinary end-fire
case is much greater than the mainlobe width for the broadside case.
104
Corectia Hansen-Woodyard
105
Hansen-Woodyard end-fire array with four
elements
𝜋 - Condition that tilt the main lobe to
𝛿 = 𝛽𝑑 −
𝑁 the right
106
• MATLAB example
N=4%number of elements
dpla=0.25%d/lambda
betad=2*pi*dpla;
fi=betad;%the phase for end-fire mode
figrd=(180*fi)/pi
i=1;
for teta = 0.0001:pi/180:2*pi
alf=betad*sin(teta)+fi;
X(i)=abs((2*sin((N*alf)/2))/(N*alf));
i =i+1;
end
teta = 0.0001:pi/180:2*pi;
polar(teta,X)%plot the AF pattern
107
- N elements BEAMSTEERED linear array
108
Beamsteering example
109
N-Elements uniform linear array directivity
when EF = 1
110
Plotting the directivity in MATLAB
113
2𝑀−1 1 1 2𝑀−1
−j 𝛽𝑑 sin 𝜃 −j 𝛽𝑑 sin 𝜃 j 𝛽𝑑 sin 𝜃 j 𝛽𝑑 sin 𝜃
𝐴𝐹𝑝𝑎𝑟 = 𝑤𝑀 𝑒 2 + ⋯+ 𝑤1 𝑒 2 + 𝑤1 𝑒 2 + ⋯ + 𝑤𝑀 𝑒 2
2𝜋
𝛽= - phase constant
𝜆
- The main idea is to avoid 𝑒 −𝑗0 term by taking the reference point in
the middle of the array
114
- Taking into account two by two left and right symmetric
terms we have:
1
−𝑗2𝛽𝑑 sin 𝜃
1
𝑗 2𝛽𝑑 sin 𝜃 1
𝑤1 𝑒 + 𝑤1 𝑒 = 𝑤1 2 cos 𝛽𝑑 sin 𝜃
2
𝑀 2𝑛 − 1
𝐴𝐹𝑒𝑣𝑒𝑛 = 2 𝑤𝑛 cos 𝛽𝑑 sin 𝜃
𝑛=1 2
- After normalization:
𝑀 2𝑛 − 1
𝑛=1 𝑤𝑛 cos 2 𝛽𝑑 sin 𝜃
𝐴𝐹𝑒𝑣𝑒𝑛 = 𝑀
𝑛=1 𝑤𝑛
115
- For an odd number of elements 𝑁 = 2𝑀 + 1
𝐴𝐹𝑜𝑑𝑑
= 𝑤𝑀+1 𝑒 −j𝑀𝛽𝑑 sin 𝜃 + ⋯ + 𝑤2 𝑒 −j𝛽𝑑 sin 𝜃 + 𝑤1 + 𝑤2 𝑒 𝑗𝛽𝑑 sin 𝜃 + + ⋯ + 𝑤𝑀 𝑒 𝑗𝑀𝛽𝑑 sin 𝜃
𝑀+1
=2 𝑤𝑛 cos 𝑛 − 1 𝛽𝑑 sin 𝜃
𝑛=1
116
- After normalization
𝑀+1
𝑛=1 𝑤𝑛 cos 𝑛 − 1 𝛽𝑑 sin 𝜃
𝐴𝐹𝑜𝑑𝑑 = 𝑀+1
𝑛=1 𝑤𝑛
117
Side lobes suppressing technique
118
PASCAL triangle
𝑛
𝑎+𝑏 = 𝑎𝑛 + +𝐶𝑛1 𝑎𝑛−1 𝑏 + 𝐶𝑛2 𝑎𝑛−2 𝑏2 + ⋯ + 𝑏𝑛
119
Example - Using side lobe suppressing technique find the weights that cancel out
the side lobes for an array with N=8.
120
•This program can be used each time one
uses sidelobes canceling criterion no matter
what function stands for the weights
generation.
121
122
• Antenna as a sensor – AoA and channel
parameter estimation
•Is about finding the angles under each front comes to the
array.
123
•Auto-correlation matrix
𝑤 𝑇 = 𝑤1 𝑤2 𝑤3 … 𝑤𝑀
𝑦 𝑘 = 𝑤 𝑇 𝑥𝑘
𝑠1 𝑘
𝑠2 𝑘
𝑥 𝑘 = 𝑎 𝜃1 𝑎 𝜃2 …𝑎 𝜃𝐷 ⋮ +𝑛 𝑘
k – sample index
𝑠𝐷 𝑘
= 𝐴𝑠 𝑘 +𝑛 𝑘
𝑎𝑇 𝜃1 = 1 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃1 𝑒 𝑗2𝛽𝑑𝑠𝑖𝑛 𝜃1
… 𝑒𝑗 𝑀−1 𝛽𝑑𝑠𝑖𝑛 𝜃1
124
•Example for three wave-fronts
1 1 1 𝑠1
𝑦 𝑘 = 𝑤 𝑇 𝑥𝑘 = 𝑤1 𝑤2 𝑤3 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃1 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃2 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃3 𝑠2
𝑒 𝑗2𝛽𝑑𝑠𝑖𝑛 𝜃1 𝑒 𝑗2𝛽𝑑𝑠𝑖𝑛 𝜃2 𝑒 𝑗2𝛽𝑑𝑠𝑖𝑛 𝜃3 𝑠3
125
𝑅𝑥𝑥 = 𝐸 𝑥 𝑥 𝐻 = 𝐸 𝐴𝑠 + 𝑛 𝑠 𝐻 𝐴𝐻 + 𝑛𝐻 - Auto-corellation matrix
𝐴 Array matrix
Is not enough !
The wave fronts must be decorelated to each other
127
Angle of arrival estimation (AoA)
• If the AoA is known at the receiver the array can cope with the frequency
selective fading using beamforming and beamsteering.
• If the AoA is determined the receiver can cancel some of the fronts placing a
NULL in its directions.
• A RADAR can determine the direction of the targets without rotating the
antenna.
The price!
The computational effort is very high
The complexity of the receiver increases
The use of signal processing techniques is a must
128
•Sensing the radio channel characteristic
129
Power delay profile
- In an urban environment
130
RMS – Delay Spread
131
Angular profile
𝑃𝑏 𝜃 = 𝑎𝐻𝜃 𝑅𝑥𝑥 𝑎 𝜃
133
134
135
136
137
• Capone method
•This method is well known as Minimum Variance Distortionless
Response. Is about finding with the maximum probability (near one) one
of the wave front, considering all others as interferers.
138
𝑤 𝑇 = 𝑤1 𝑤2 𝑤3 … 𝑤𝑀
−1
𝑅𝑥𝑥 𝑎 𝜃
𝑤= 𝐻 - Calculate the weights for
𝑎 𝜃 𝑅𝑥𝑥 𝑎 𝜃 direction theta to maximize the
power for that direction and
minimize the power for others
139
140
Finding the weights
141
• Linear prediction method
The main purpose of this method is to minimized the error between one
element and a specific one that is the reference for all others.
𝑢4𝑇 = 0 0 0 1 0 0 … . . 0
𝑇 𝑅 −1 𝑢
𝑢𝑚 𝑥𝑥 𝑚
𝑃𝑃𝐿𝑚 𝜃 = 2
𝑇 −1
𝑢𝑚 𝑅𝑥𝑥 𝑎 𝜃
142
143
• Burg method
1
𝑃𝐵𝑀𝑒𝑗 =
𝑎𝐻𝜃 𝑐𝑗 𝑐𝑗𝐻 𝑎 𝜃
InvRx=inv(Rxx);
cj=InvRx(:,m)%genereaza coloana j
144
145
146
147
148
149
• Pisarenko method (harmonic decomposition)
- Is about the minimization of the mean square error taking into account
that the noise is decorelated with the direction from which the wave
fronts coming from. According to Pisarenko the estimation probability of
the incoming wave front under theta is:
1
𝑃𝑃𝐻𝐷 𝜃 =
𝑎𝐻 𝜃 𝑒1 2
150
151
152
153
154
155
•Observatie
156
MUSIC – (MULTIPLE SIGNAL CLASIFICATION)
- Euclidian distance!!!
Remark!
That the array vector for ULA configuration is a polynomial of degree M-1
- Diagonal elements
Multiplication
with
- 2(M-1) roots
- Our solutions must be complex with the magnitude near unity!
- These are the poles located near the trigonometric unity circle
- M=4, element spacing = 0.5 wavelength, noise variance 0.1 with source
correlation. AoA=[-4 4].
- Doublets concept
- Invariant concept
It’s true!
- ESPRIT – EXAMPLE
- M=4, element spacing= 0.5 wavelength, noise variance 0.1 with
source correlation. AoA=[-2.5 2.5]. Training set uses K=1325
samples.
Remark!
In order to compare ESPRIT with MUSIC in the same conditions we
have to take M = doublet dimension.
Why?
Antene pentru microunde - Nicolae 174
Crisan
Antene pentru microunde - Nicolae 175
Crisan
Ec –dimension 2Dx2D
- ESPRIT
- Remarks
- ESPRIT -> both front are trapped
AoA=[-2.5 2.5] with source correlation with an error of about 0.2 degree
and K=1350 - MUSIC -> has only one SVD
matrix decomposition
- MUSIC - The accuracy is almost the same
•In IF domain the computational effort is important but leads to the best results.
179
3.1. RF domain array processing
• The adaptation methods in RF domain are very effective and cheap for
beamforming. For example the most simple and straight RF domain processing
is the side lobes suppressing method (with real weights).
• Further only the RF domain processing methods with complex weights will be
studied.
180
181
The array factor will be:
3 1 1 3
−𝑗 𝛽𝑑𝑠𝑖𝑛(𝜃 −𝑗 𝛽𝑑𝑠𝑖𝑛(𝜃 𝑗 𝛽𝑑𝑠𝑖𝑛(𝜃 𝑗 𝛽𝑑𝑠𝑖𝑛(𝜃
𝐴𝐹𝑘 = 𝑤𝑘1 𝑒 2 + 𝑤𝑘2 𝑒 2 + 𝑤𝑘3 𝑒 2 + 𝑤𝑘4 𝑒 2
3𝜋 𝜋 𝜋
−𝑗 4 −𝑗 4 −𝑗 2
𝑤𝑘1 = 𝑒 ; 𝑤𝑘2 = 𝑒 −𝑗𝜋 ; 𝑤𝑘1 = 𝑒 ; 𝑤𝑘1 = 𝑒
182
183
There are many possibilities to achieved beamforming by change the
position of the -45 degree delay line or its value.
184
Antene pentru microunde - Nicolae 185
Crisan
- Hybrid quadrature coupler can be simulated in HFSS, ADS or
SysVue in order to counteract the effect of losses in dielectric and
balance the paths against the imbalance effect of the mutual coupling
between the lines.
186
187
188
189
•Smart beamforming
Both of them are demanding a pre-processing step in which case the auto-
correlation matrix must be captured.
The main difference is related to the fact that the former proceeds in time domain
and the latter in frequency domain.
190
191
- ABF uses transmition lines
- DBF uses DSP’s
𝐴𝑜𝐴 = [𝜃1 𝜃2 𝜃3 … . 𝜃𝐷 ]
193
𝑤1 + 𝑤2 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃1
+ ⋯ + 𝑤𝑀 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃1
=1
𝑤1 + 𝑤2 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃2
+ ⋯ + 𝑤𝑀 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃2
=0
…
𝑤1 + 𝑤2 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃𝐷
+ ⋯ + 𝑤𝑀 𝑒 𝑗𝛽𝑑𝑠𝑖𝑛 𝜃𝐷
=0
𝑥1 + 𝑎1 𝑥2 + ⋯ + 𝑎M−1 𝑥M = 1
𝑥1 + 𝑏1 𝑥2 + ⋯ + 𝑏M−1 𝑥M = 0 Zero Forcing algorithm (in LTE
…
𝑥1 + 𝑐1 𝑥2 + ⋯ + 𝑐M−1 𝑥M = 0 standard)
•If 𝐴−1 does not exist we must increase M or modify the array
spacing
195
196
197
198
199
200
• MMSE (Minimum mean-square error) beamforming
•The main idea here is to minimize the MMSE error as much as possible
201
- 𝑑(𝑘) is the reference sample and must be decorelated with the
interference signal and correlated with selected signal 𝑠𝑖
202
- The optimal weights can be found as:
where:
- 𝑆𝑠(𝑘) is the normalized energy of selected
wave front,
- 𝑅𝑥𝑥 , correlation matrix at the receiver
(MxM)
- 𝑎(𝜃𝑠 ) array vector along direction under 𝜃𝑠
angle
203
204
205
206
• SIR maximization
𝑅𝑠𝑠 = 𝐸 𝑥𝑠 𝑥𝑠𝐻
207
𝑦 𝑘 = 𝑤𝐻 𝑥 𝑘 = 𝑤𝐻 𝑥𝑠 𝑘 + 𝑥𝑖 𝑘 +𝑛 𝑘 = 𝑤𝐻 𝑥𝑠 𝑘 +𝑢 𝑘
- 𝑘 current sample,
- 𝑢 𝑘 all interference signals with noise
- 𝑛 𝑘 noise for k sample
- 𝑤𝐻 complex weights
𝑅𝑠𝑠 = 𝐸 𝑥𝑠 𝑥𝑠𝐻
−1
𝑅𝑢𝑢 𝑅𝑠𝑠 𝑤𝑆𝐼𝑅 - Where 𝑤𝑆𝐼𝑅 is the eigenvector of
𝑤𝑜𝑝𝑡 = −1 𝑅 that corresponds
the matrix 𝑅𝑢𝑢
𝜆𝑚𝑎𝑥 𝑠𝑠
to the maximum eigenvalue 𝜆𝑚𝑎𝑥
208
209
210
211
212
213
214
Array antenna over MIMO channels
215
MIMO – Beam forming using complex weight coefficients
216
Receiver-Combining Methods
ℎ1
ℎ2
ℎ3 𝑦 4
𝐸𝑠 2
𝛾= ℎ𝑖
ℎ4 𝑁
𝑖=1
ℎ1 𝑥1 ℎ1 𝑦1 𝑥1 ℎ1
ℎ 𝑥1 ℎ2 𝑦 𝑥1 ℎ2
𝑦 = 𝐻 𝑡 𝑥1 = 2 𝑥1 = = 𝑦2 𝑦=𝐻 ∗ 𝑦 = ℎ1∗ ℎ2∗ ℎ3∗ ℎ4∗
ℎ3 𝑥1 ℎ3 3 𝑥1 ℎ3
ℎ4 𝑥1 ℎ4 𝑦4
𝑥1 ℎ4
217
Autocorrelation matrix – LTE channel model
218
LTE – structure in time domain
219
220
LTE – Resource grid in frequency domain
221
Array antenna – channel matrix estimation
222
Channel estimation 2x2 MIMO
n – sample index
223
224
Transmit antenna diversity 2x2 mod 2 LTE - Alamouti
225
Transmit antenna diversity 2x2, 4x4 mod 2 LTE
226
Alamouti - decoder
but
- Pairs of
consecutive
sequences
- When channel is
assumed to be static
during n and n+1
227
Rank-Deficiency Problem
- X transmitted signal
- H channel matrix
- Y received signal
- n the noise
𝐻 ≠ 0 ; is a must
- X transmitted signal
- H channel matrix
- V precoding matrix
- Y received signal
- n the noise
𝐺𝑀𝑀𝑆𝐸 = 𝐻 𝐻 𝐻𝐻 𝐻 + 𝜎𝑛2 𝐼𝑀 −1
𝑤𝐻 = 𝑢𝑖𝑇 𝐴𝐻 𝐴𝐴𝐻 + 𝜎𝑛2 𝐼𝑀 −1
ℎ11
ℎ11 ℎ12 ℎ13 ℎ14
ℎ ℎ22 ℎ23 ℎ24
𝐻 = 21
ℎ31 ℎ32 ℎ33 ℎ34 ℎ22
ℎ41 ℎ42 ℎ43 ℎ44
ℎ33
ℎ44
𝐻 = 𝐴𝐻𝑠
ℎ11 2 0
ℎ11 0 0 0 0 0
2 0 0
0 ℎ22 0 0 0 ℎ22
𝐻𝑠 = 𝑅𝑠𝑠 = 2
0 0 ℎ33 0 0 0 ℎ33 0
0 0 0 ℎ44 2
0 0 0 ℎ44