25
Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Embed Size (px)

Citation preview

Page 1: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Decoupling FeedingNetwork for

Antenna ArraysStudent: Eli RivkinSupervisor: Prof. Reuven ShavitDepartment of Electrical and Computer Engineering

BGU

Page 2: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Table of Contents BGU

Motivation – mutual coupling and its effect on antenna arrays

Eigenmode theoryDecoupling conceptHardware implementation of the

decoupling networkConclusions

Page 3: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Introduction

One of the major problems in antenna arrays is mutual coupling among the elements.

BGU

Page 4: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

So What’s the Problem?

Single antenna:

Antenna + friend:

BGU

Page 5: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

More Specifically - What’s the Problem with the Mutual

Coupling Again?• Difficulties in designing a predefined radiation pattern.• Gain reduction, especially at scanning.• Power mismatch => losses, reflections.• Matching is possible only for one excitation, not

always.

BGU

Page 6: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Main GoalTo design a feeding network connected to the antenna array so that its input ports are always matched, independently of the mutual coupling.

0inS

Decoupling &Matching

Network(DMN)

Antenna Array

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

BGU

Page 7: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Geometry

x

z

d

d

d

d

d

d(4,4)

(4,3)

(4,2)

(3,4)

(3,3)

(3,2)

(2,4)

(2,3)

(2,2)

(1,4)

(1,3)

(1,2)

(3,1)

(2,1)

(1,1)

(4,1)

PEC

y

dipole2

BGU

V1,1

I1,1

V1,2

I1,2

.

.

.

h

z

y

PEC

x

(1,1)

(1,2)

Page 8: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Mutual Coupling1 11 1 12 2 1

2 21 1 22 2 2

1 1 2 2

...

...

.................................

...

M M

M M

M M M MM M

V Z I Z I Z I

V Z I Z I Z I

V Z I Z I Z I

1 11 2

,1 ,2 , 1 , , 1 ,... ...a i i i Mi i i i i i i i i i M

i i i i i i

V I II I IZ Z Z Z Z Z Z

I I I I I I

Input impedance at each port )active impedance(:

depends on the excitation!

BGU

1I Z V Y V

11 1

1

M

M MM

Y Y

Y

Y Y

- admittance matrix

1 11 1 12 2 1

1 1 2 2

...

.................................

...

M M

M M M MM M

I Y V Y V Y V

I Y V Y V Y V

11 1

1

M

M MM

Z Z

Z

Z Z

- impedance matrix

V Z I

each element’s current depends on the voltages of all the others!

V1

I1

V2

I2

.

.

.

Page 9: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Radiation Pattern

( , )F

4 4

( 1)sin cos ( 1)cos

1 1

( , ) jkd m nmn

m n

AF I e

E-plane: 90

H-plane: 90

AF (array factor)AF (array factor)

4 4

( 1)sin cos ( 1)cos

1 1

jkd m nmn

m n

I e

EF (element factor)EF (element factor)

cos cos sin sin sin2 2

sin

x

z

d

d

d

d

d

d

PEC

y

r

(x,y,z)

BGU

Page 10: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Examples

1. Difference pattern )Bayliss(2. Sum pattern )Taylor(

BGU

zoom:

Page 11: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Eigenmode Theory[ ]b S aH H

i rP a a P b b

[ ] [ ] [ ]H H H H Hrad i rP P P a a b b a U S S a a H a

[H] is Hermitian )[H]H=[H]( => it can be diagonalized by a unitary matrix:

[ ][ ][ ]HH Q Q

where:

[ ][ ] [ ]HQ Q U )unitary( 1 2{ }

H

Mdiag Q H Q )diagonal(

The columns of [Q] are the eigenmodes of the antenna array. They are orthonormal vectors.0 1i - eigenefficiencies

(eigenvalues)

BGU

[H] - radiation matrix

Antenna Array

[S]

b

a

Page 12: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Eigenmode Theory )cont’d(

If [Q] diagonalizes [H] )as shown previously( then it also diagonalizes [S] via:

*[ ] [ ][ ]HS Q Q

where:

1 2{ }T

Mdiag Q S Q )diagonal(

i- modal reflection coefficients )complex, (- modal reflection coefficients )complex, (0 1i

21i i => energy conservation

[ ][ ][ ]HH Q Q

[ ] [ ] [ ]HH U S S - radiation matrix

BGU

Page 13: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Decoupling Concept

bs

as Antenna Array

[S]

Decoupling Network

[SD]

b

a

BGU

11 1

1

M

M MM

S S

S

S S

,21

,21

0T

D

D

D

SS

S

,21 ,21[ ]T

S D DS S S S

from the theory: 1

2

0 0

0

0

0 0

T

M

Q S Q

SS

,21[ ] [ ]DS Q

SS

1

2

0 0

0

0

0 0 M

Page 14: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Decoupling Concept - Conclusions

• It is reciprocal and lossless.• Its input and output ports are matched and

decoupled.• Power is transferred according to the matrix of

eigenmodes [Q].• Each input port excites a different eigenmode =>

every excitation is a superposition of the orthonormal eigenmodes.

• All the input ports are independent of each other, so now it is possible to match each port individually.

0

T

D

QS

Q

• The decoupling network is described by: ]( Q -[matrix of eigenmodes)

this network matches the system regardless of the excitation!

BGU

Page 15: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Numerical Results BGU

z

x1234

5678

9101112

13141516difference pattern)Bayliss(

sum pattern)Bayliss(

Page 16: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN

BGU

Objective: implementation of with passive microwave elements. 32 32

0

0

T

D x

QS

Q

Special case: array of 2 antennas with , so that: .11 22s s 11 12

2 212 11

x

s sS

s s

In this case, will diagonalize [S] and decouple the 2-element array. 1 1

1 12

jQ

0 0 1 1

0 0 1 1

1 1 0 02

1 1 0 0

D

jS

Directional Coupler )Magic-T Hybrid(

Our case: symmetric rearrangement of the elements in [S] leads to:

11 128 8 8 8

16 1612 118 8 8 8

x xsym x

x x

S SS

S S

According to the special case, our [Q] can be written in block matrix notation:

(1) 8 8 8 8

16 168 8 8 82x x

xx x

U UjQ

U U

8 Magic-T Hybrids

8 8 8 816 16

8 8 8 8(1)

32 328 8 8 8

16 168 8 8 8

0

20

x xx

x xD x

x xx

x x

U U

U UjS

U U

U U

#1

#2

#3

#4

Page 17: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

So what does “symmetric rearrangement of the elements in [S]” mean?z

x1234

5678

9101112

13141516

symmetryplane

symmetric rearrangement

z

x1234

5678

9101112

13141516

original arrangement

pairs: 1 & 9, 2 & 10 …

Each of these pairs should be connected by a coupler.

After this step the system matrix will be:

12 11(1) (1) (1) 8 8 8 8

16 1616 16 16 16 16 1612 118 8 8 8

0

0

Tx x

S sym xx x xx x

S SS Q S Q

S S

no coupling between the 2 groups!

Page 18: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

How will it look like so far?

(1)

32 32D xS

#9

#1#1

#2

#3

#4

#1'

#9'

#10

#2#2'

#10'

#16

#8#8'

#16'

#1

#2

#3

#4

#1

#2

#3

#4

.

.

.

11 128 8 8 8

16 1612 118 8 8 8

x xsym x

x x

S SS

S S

12 11(1) 8 8 8 8

16 1612 118 8 8 8

0

0x x

S xx x

S SS

S S

Page 19: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

z

x1234

5678

9101112

13141516

z

x1265

3487

11121615

9101413

symmetryplane

What’s next?

• Symmetric rearrangement of [SS)1(] )division of each group of

8 into 2 symmetric groups of 4 elements(.• Connection of 8 more Magic-T Hybrids )4 for each group(.• After this there will be no coupling between the 4 groups of 4.

Same procedure as before:

Page 20: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

How will it look like so far?

1 4 44 4

8 824 4 4 4(2)

16 163 4 44 4

8 844 4 4 4

00

0

00

0

xx

x

x xS x

xx

x

x x

S

SS

S

S

#1

#2

#3

#4

#1''

#5''

#4''

#8''

#1’

#5’

#4’

#8’

#1

#2

#3

#4

#12''

#16''

#12’

#16’

#1

#2

#3

#4

.

.

.

(2)

32 32D xS

#9''

#13''

#9’

#13’

#1

#2

#3

#4

.

.

.

(1)

32 32D xS

#9

#1#1

#2

#3

#4

#1'

#9'

#16

#8#8'

#16'

#1

#2

#3

#4

.

.

.

#12

#4#1

#2

#3

#4

#4'

#12'

#13

#5#5'

#13'

#1

#2

#3

#4.

.

.

12 11(1) 8 8 8 8

16 1612 118 8 8 8

0

0x x

S xx x

S SS

S S

Page 21: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

So what do we have so far?• 4 independent sub-arrays of 4 elements each.• Each sub-array has known [S] and [Z] matrices, calculated

in MATLAB.• No more symmetry planes have left, so it’s impossible to

use the same method again.

A different method will be used to decouple each of the sub-arrays.The method is based on diagonalizing the imaginary and the real partsseparately.

[S]...

[SA]...

in inZ or S Z or S

columns of [A] are orthonormal realvectors, then besides ,also: .

0

0

T

A

AS

A

T

inS A S A

T

inZ A Z A

Theorem: if , and the

Page 22: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

[S1]

-jx1

-jx2

-jx3

-jx4

[SA]

0

0

T

A

AS

A

[SB]

0

0

T

B

BS

B

1 1 1Z R j X

1

21 1 1 1

3

4

0 0 0

0 0 0'

0 0 0

0 0 0

T T T

A

x

xZ A Z A A R A j A R A j

x

x

1 1 1' ' 'Z R j X

The columns of [A] are the eigenvectors of [X1]. Since they are real andorthonormal, the theorem can be used:

1 '''Z

1 1 1''T T

A AZ A R A j j A R A

1 1'T

AZ A R A j

1 1''T

Z A R A

The columns of [B] are the eigenvectors of [A]T[R1][A]. Since they are real and orthonormal, the theorem can be used again:

1

21 1

3

4

0 0 0

0 0 0'''

0 0 0

0 0 0

T T

B

r

rZ B A R A B

r

r

B

the input impedance matrix is diagonal -decoupling accomplished!

Page 23: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

Hardware Implementationof the DMN )cont’d(

BGU

Last thing left to do is to implement [SA] and [SB].

4 4 4 48 8

4 4 4 4

0

0

T

x xA x

x x

AS

A

Using Givens Rotations, [A] can be expressed as:where each one of [Ai] is a matrix which represents a directional coupler )an arbitrary one, not Magic-T as before(.

1 2 6...A A A A

[SA] and [SB] are implemented by 6 cascaded couplers each.

Decoupling a 4-element sub-array requires 12 couplers.

The 4 sub-arrays which were left after the first method require 48 couplers.

Page 24: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

ConclusionsBGU

The suggested decoupling network achieves the goal- the system will be always matched.

It can be implemented both in software and hardware.

Software implementation requires connecting the antenna array to a computer which does all the matrix calculations )after translating the signal to baseband and sampling(.

Hardware implementation requires 64 directional couplers )16 for the first steps with the first method, 48 for the last step with the second method(. All the parameters of the couplers were calculated in MATLAB.

Using only the second method, which is a general one )not depending on symmetries(, would require 320 couplers.

Page 25: Decoupling Feeding Network for Antenna Arrays Student: Eli Rivkin Supervisor: Prof. Reuven Shavit Department of Electrical and Computer Engineering BGU

ReferencesBGU

[1] Volmer, C., Weber, J., Stephan, R., Blau, K. and Hein, M.A., "An Eigen-Analysis of Compact Antenna Arrays and Its Application to Port Decoupling", IEEE Transactions on Antennas and Propagation, Vol. 56, No. 2, February 2008.

[2] Geren, W.P., Curry, C.R. and Andersen, J., "A Practical Technique for Designing Multiport Coupling Networks", IEEE Transactions on Microwave and Techniques, Vol. 44, No. 3, March 1996.