Investigation of High Resolution DOA
Estimation Algorithms for Optimal
Performance of Smart Antenna
Systems
(M.Sc. by research – Preliminary results)
By: Ebrahim Al Ardi
(Etisalat Telecommunication Corporation)
1
Outlines
•
What is a Smart Antenna ?
•
Advantages of deploying Smart Antennas
•
DOA Estimation Algorithms
•
Simulation Results and Numerical Experiments
•
Conclusions
2
1
What is a Smart Antenna ?
Smart
Not Smart
• Array of antennas (Linear, Circular, Planar)
• DSP (Digital Signalling Processing)
(1) Direction Estimation
(2) Adaptive Beam- forming
3
Advantages of deploying Smart Antennas
• Enhancement of coverage range
• More power efficient
• Improvement of system capacity
• Increasing the chances for the application of new services
4
2
DOA Estimation Algorithms
MUSIC Algorithm:
• High-resolution algorithm
• Produces its output as spectrum of power
• Time consuming
Root MUSIC Algorithm:
• Based on polynomial roots solving
• Applicable for Uniform Linear Arrays only
ESPRIT Algorithm:
• Reduces the computational time compared to MUSIC
• Based on dividing the sensor array into sub-arrays
5
Simulation Results and Numerical Experiments
(1)
Parameters that affect the performance of the DOA Algorithms:
• Spacing between the array elements
• Angular separation between the incident signals
• Number of samples taken for the incident signals
• Signal-to-Noise Ratio
• Number of array elements
6
3
Simulation Results and Numerical Experiments
(2)
Effect of increasing the spacing between the array
elements
180
N=2
K = 50
SNR = 20
160
Detected angle in Degrees
140
120
100
80
60
40
20
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Element spacing normalized to wavelength λ
1.8
2
7
Simulation Results and Numerical Experiments
(3)
Effect of increasing the spacing between the array
elements
0
0
N = 2 K = 50 SNR = 20 d = 0.4 λ
N = 2 K = 50 SNR = 20 d = 0.8 λ
-20
-30
-40
-50
-60
desired peak
-10
Relative Power (in dB)
Relative Power (in dB)
-10
unwanted peak
-20
-30
-40
-50
0
20
40
60
80
100
120
Angles of Arrival in degrees
140
160
180
-60
0
20
40
60
80
100
120
Angles of Arrival in degrees
140
160
180
8
4
Simulation Results and Numerical Experiments
(4)
Effect of increasing the angular separation between
the incident signals
0
0
N = 4 K = 50 SNR = 20 d = 0.5 λ
N = 4 K = 50 SNR = 20 d = 0.5 λ
-5
-10
-15
Relative Power (in dB)
Relative Power (in dB)
-10
-20
-25
-30
-20
-30
-40
-35
-50
-40
-45
0
20
40
60
80
100
120
Angles of Arrival in degrees
140
160
180
(a) Incident angles = 40o and 60o
-60
0
20
40
60
80
100
120
Angles of Arrival in degrees
140
160
(b) Incident angles = 20o and 80o
Simulation Results and Numerical Experiments
180
9
(5)
Effect of increasing the angular separation between
the incident signals
φ Exact
φ ESPRIT
∆φ
40o
60o
40.58 o
60.47 o
0.58 o
0.47o
20o
80o
19.99 o
80.05 o
0.01o
0.05o
10
5
Simulation Results and Numerical Experiments
(6)
Effect of increasing the number of samples taken for
the incident signals
K
Radial
amplitudes of
the roots
Radial distance
to the unit circle
50
1.06
1.03
0.92
0.06
0.03
0.08
500
1.01
1.01
0.98
0.01
0.01
0.02
11
Simulation Results and Numerical Experiments
(7)
Effect of increasing the number of array elements and SNR
The effect of increasing the number of array elements
and Signal-to-Noise (SNR) ratio has also been investigated
and it was found that the performance of the algorithms
improves when more elements are used and when the
SNR is increased
12
6
Simulation Results and Numerical Experiments
(8)
Computational Time of the Algorithms
Execution time in seconds
2.5
MUSIC
Root MUSIC
ESPRIT
2
1.5
1
0.5
5
10
15
20
25
30
35
40
No. of array elements
45
50
55
60
13
Conclusions
•
ETISALAT participates in, and encourages, the research efforts by
establishing programs that concentrate in the research and evaluation of
new challenging services and technologies
•
Performance evaluation of high-resolution DOA estimation
algorithms including MUSIC, Root-MUSIC and ESPIRIT
has been carried out
•
Performance of the three algorithms improves by using more
elements in the sensor array and more samples of the incident
signals, as well as increasing the signal-to-noise ratio and angular
separation between the incident signals.
14
7
Conclusions
•
Different simulations showed that MUSIC algorithm takes more time
to produce accurate results when fewer elements are used in the
sensor array. However, as the number of array elements increases,
the computational time of ESPRIT algorithm starts to increase drastically
especially when a large number of samples of the incident signals is
used.
•
Using element spacing for the sensor array up to 0.5λ allows the
detection of all possible angles of incidence. However, using larger
values for the element spacing will lead to the detection of signals
arriving only at larger angles of incidence (i.e., approaching normal
incidence).
15
Thanks
Ebrahim Al Ardi
Engineer/Mobile Systems/Quality Assurance
Emirates Telecommunications Corporation (ETISALAT)
Tel: +971-4-2022922, Fax: +971-4-2210060, mobile: +971-50-6623963
Email: ardi@emirates.net.ae
16
8