Академический Документы
Профессиональный Документы
Культура Документы
MaxSignal
MaxGain
LCMV Constraind
30.
10
LCMV based
Algorithm
-
Loop: Thinning
LCMV Beamformer
Remove Min weights, Optionally:
Symmetrical
Fixed Extremes
Test Case
-
Radiators: Isotropic
Scanning: Broadside; SLL: 13db, -25db
Size: 20 elements, lambda/2 distance
Results
Algorithm tend to simply remove edge elements, effectively synthetizing a smaller
array with original inter element distance.
This effect is shown also if edge elements are kept always active, es:
(10000011111111000001)
Considerations
SLL constraints on patter cause a resulting tapering in the weights, resulting in weak
edge elements. Therefore algorithm tends to iteratively remove extremes, resulting
in simply a smaller array.
Directions
It is clear that non-tapered weighting should be synthetized for this approach to be
effective.
To obtain so the MaxGain beam former is proposed for obtaining a more suitable
weighing vector.
Max signal Beamformer is not applicable as this return equiamplitude coefficients
and no minima would be present.
01.
11
Algorithm
-
Loop: Thinning
MaxGain Breamformer
Remove Min weights symmetrically
Test Case
-
Radiators: Isotropic
Scanning: Broadside, 45
Size: 20 elements, lambda/2 distance
Results
Sparsification is achieved for 45:
-
DEL:019
DEL:017
DEL:007
DEL:011
DEL:008
>
>
>
>
>
10111111111111111101
10101111111111110101
10101101111110110101
10101101100110110101
10101100100100110101
Loop: Thinning
MaxGain Beamformer
Remove min weights (not symmetrical)
Test Case
-
Results
For cos^n type radiation type results show high sparsification.
For broadside scanning the obtained solutions are the same as the optimal ones
(combinatorialy testing).
For off-broadside scanning (20) optimal solutions are still achieved until a critical
thinning factor.
After this the optimal solution have a drastically different layout, resuming element
previously off.
This algorithm, proceeding by iteratively removing element, is not able to synthetize
this.
For high embedded element tapering coefficient (n) weighing vector coefficient
appears to oscillate, resulting probably in less effective weighing vector (though not
extensively tested).
This effect is recorded also for growing scanning direction.
Element number does not affect the global thinning solution as the envelope of the
weighting vector is maintained for very large and small array.
Considerations
As expected, algorithm is completely ineffective for Isotropic radiators.
Directions
Algorithm is to be improved with SLL constrains and enhanced to be able to restore
elements.
16.
11
Algorithm
-
Loop:
Thinning
MaxGain Beamformer
LCMV Beamformer
Remove min weights (not symmetrical)
Test Case
Testing case is chosen to be similar to realistic setup for RUAG
-
Radiators: cos^n
Scanning: Broadside; SLL: -27
Size: 100-200 elements, distance from HPBW+Scanning: eg 40,16 ~ 1.5
lambda
Results
Array is effectively sparsified, though given the large array size is difficult to
compare with optimal solution.
13.
11
MaxGain with T
Algorithm
-
Loop: Thinning
Loop: Convergence
MaxGain Beamformer (TA)
T=1/(w+e)
Remove min weights (not symmetrical)
Test Case
Testing case is chosen to be similar to realistic setup for RUAG
-
Radiators: cos^n
Scanning: Broadside; SLL: -27
Size: 100-200 elements, distance from HPBW+Scanning: eg 40,16 ~ 1.5
lambda
Results
Array is effectively sparsified, though given the large array size is difficult to
compare with optimal solution.
13.
11
LCMV with T
Algorithm
-
Loop: Thinning
Loop: Convergence
LCMV Beamformer (TA)
T=1/(w+e)
Remove min weights (not symmetrical)
Test Case
Testing case is chosen to be similar to realistic setup for RUAG
-
Radiators: cos^n
Scanning: Broadside; SLL: -27
Size: 100-200 elements, distance from HPBW+Scanning: eg 40,16 ~ 1.5
lambda
Results
Array is effectively sparsified, though given the large array size is difficult to
compare with optimal solution.