Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Remote Active Noise Control
1. Remote FxLMS Algorithm for Active Control
of Sound in Remote Locations
Iman Ardekani
Department of Computing
Unitec Institute of Technology
Auckland, New Zealand
Waleed Abdulla
ECE Department
The University of AUckland
Auckland, New Zealand
APSIPA ASC 2014
APSIPA Annual Summit and Conference
Cambodia, Dec. 9 – 12, 2014
3. Active Noise Control – Why?
𝜆
𝜆 =
𝑐
𝑓
wave length (m) sound velocity (m/s)
frequency (Hz)
𝑑
effective passive control 𝑑 ≫ 𝜆
𝑓 (𝐻𝑧) 𝜆 (𝑚)
10000 0.0343
1000 0.343
100 3.343
Passive noise control is
bulky and costly for low
frequencies!
3
4. Active Noise Control – Original Idea
4
Primary Noise
Secondary Noise
Residual noise
5. Active Noise Control – Acoustic Domain
5
u(n) : original noise
d(n) : primary noise
Reference
mic
Error
mic
d’(n) : secondary noise
v(n) : anti noise d(n) u(n)
d’(n) v(n)
e(n)
u(n) v(n)
e(n)
Control
System
Control
System
6. Active Noise Control – Digital Electronic Domain
6
W
Gu(n) e(n)
FxLMS
H
v(n)
Control System
d(n)
d'(n)
Minimization of e(n) power through producing v(n) using u(n) and e(n)
FxLMS
Algorithm
7. Active Noise Control – Research Gap
7
𝜆
20
Traditional ANC
e(n)
Problems:
- very small zone of quiet
- space occupied by the error mic
10 dB ZoQ
𝜆
20
e(n)
10 dB ZoQ
Advantage:
- effective use of space in quiet zone
Remote ANC (proposed)
8. ANC Analysis in Acoustic Domain – Coordinate System
8
u(n)
Reference mic
v(n)
L1
e(n)
Control Source
L2 Lo
Error mic (ZoQ centre)
W
G e(n)
FxLMS
Update Eq
H
+
v(n)
Digital control domain
Physical domain
z-axis
L1
L0 L2
is
9. ANC Analysis in Acoustic Domain – Propagation
9
L1
z-axis
y-axisϕ
1
d
D
L1
L2L0
z-axis
y-axisϕ
1
d
D
L1
L2L0
ϕr
Lr
ϕr
u(n)
v(n)
W
Gu(n) e(n)
FxLMS
H
+
v(n)
Digital control domain
Physical domain
10. Remote Active Noise Control
10
Remote location
D
L1
z-axis
y-axisϕ
1
d
D
L1
L2L0
ϕr
Lr
ϕr
11. Remote Active Noise Control
11 L1
z-axis
y-axisϕ
1
d
D
L1
L2L0
z-axis
y-axisϕ
1
d
D
L1
L2L0
ϕr
Lr
ϕr
12. Remote Active Noise Control
12
L1
L0 L2
z-axis
y-axisϕ
1
d
D
L1
L2L0
z-axis
y-axisϕ
1
d
D
L2L0
ϕr
Lr
ϕr
y-axisϕ
1
d
D
L1
L2L0
z-axis
y-axisϕ
1
d
D
L1
L2L0
ϕr
Lr
ϕr
Traditional ANC Remote ANC
13. Remote Active Noise Control
13
Traditional ANC Remote ANC
W
Gu(n) e(n)
FxLMS
H
+
v(n)
Digital control domain
Physical domain
z-axis
L1
L0 L2
z-axis
Krz
Gu(n) e(n)
Remote
FxLMS
H
v(n)
W
a(n) -ϕr
Digital control domain
+
Physical domain
14. Remote Active Noise Control
14
Traditional ANC Remote ANC
W
Gu(n) e(n)
FxLMS
H
+
v(n)
Digital control domain
Physical domain
z-axis
L1
L0 L2
z-axis
Krz
Gu(n) e(n)
Remote
FxLMS
H
v(n)
W
a(n) -ϕr
Digital control domain
+
Physical domain
18. Conclusion
• A novel model for the analysis of the ANC systems in
the acoustic domain is proposed.
• Based on this model, a methodology for active noise
control in remote location is developed.
• An adaptive framework for the realization of the
proposed remote ANC system is developed.
• Using remote ANC idea, the space available in the
quiet zone can be used more effectively.
• Future work: targeting 3D zones of quiet in remote
locations instead of a point.
Page 18
𝜆
20
e(n)
10 dB ZoQ
Editor's Notes
In practice, we have to place a microphone close to the noise source in order to estimate the original noise signal. However, this measurement is not enough for producing an accurate anti-noise signal. This is because the noise received at the desired quiet zone is slightly different from the original noise. This difference is due to the influences of electro-acoustic signal channel between the noise source and the desired quiet zone.
For solving this problem we have to place a microphone in the quiet zone. We call this microphone error microphone. This microphone cannot measure the primary noise because the primary noise is intended to be combined by the secondary noise at the quiet zone. However, this microphone can measure the residual noise that is the combination of the primary and secondary noise signals at the quiet zone.
The information provided by the error microphone along with the information provided by the reference microphone is used by the control system to estimate and produce an optimal anti-noise signal.
In practice, we have to place a microphone close to the noise source in order to estimate the original noise signal. However, this measurement is not enough for producing an accurate anti-noise signal. This is because the noise received at the desired quiet zone is slightly different from the original noise. This difference is due to the influences of electro-acoustic signal channel between the noise source and the desired quiet zone.
For solving this problem we have to place a microphone in the quiet zone. We call this microphone error microphone. This microphone cannot measure the primary noise because the primary noise is intended to be combined by the secondary noise at the quiet zone. However, this microphone can measure the residual noise that is the combination of the primary and secondary noise signals at the quiet zone.
The information provided by the error microphone along with the information provided by the reference microphone is used by the control system to estimate and produce an optimal anti-noise signal.
Small size of the quiet zone
Shows the validity of our analysis in the acoustic domain
Small size of the quiet zone
Shows the validity of our analysis in the acoustic domain