Telecoms


ACOUSTIC CHANNEL EQUALISERS FOR ACTIVE NOISE CONTROL APPLICATIONS

Research activity carried out by Dragana Nikolic and Dr Branislav Vuksanovic in collaboration with ECASS Technologies Ltd


BACKGROUND

Active noise control is the process of reducing unwanted sound by combining it with a sound of the same amplitude but of opposite phase. The approach investigated in this work is to cancel sound in unconfined spaces (outdoors), where the size of the quiet zone (acoustic shadow) is not restricted to within an acoustic wavelength. Produced quiet regions can therefore extend to large distances.

The proposed ANC system uses an active sound barrier of secondary sources to cancel the unwanted sound from the primary source at an array of error microphones. By cancelling the sound at the error microphones distributed across the controlled region, the secondary sources create a zone of reduced noise over this area.

ANC system scheme


ANC system scheme

Most ANC systems use transverse FIR filters together with LMS and FXLMS algorithms to cancel both predictable (periodic) and unpredictable (broadband) types of acoustic noise. The cancelling structures used to implement those algorithms require considerable computational processing effort to adapt to the primary source and plant changes, particularly for multichannel (MIMO) systems, where the number of control loops increases according to the square of the number of channels. Therefore, to implement an efficient and fast response to source changes, including unpredictable, statistically non-stationary noise; large online adaptive FIR filters need to be dispensed with. A time domain solution that gives virtually instantaneous cancellation response to primary source changes, and is computationally efficient, is to use a copy of the primary source signal, compensated for plant distortion by passing the signal through the plant inverse, aligned at the position where the secondary wave coincides with the primary wave and matched in amplitude, is used directly to cancel the sound.

This approach is expected to reduce computational requirements of the system and to increase the response speed of the MIMO system for cancelling the stationary random sound. A MIMO ANC system needs not only to estimate inverses of the direct paths but also to somehow invert the cross paths in the system and to include those in a multichannel set-up. Figure below shows a possible set-up of a two-channel ANC system, which does account for the influence of the acoustic cross paths in the system. Here, cross coupling of the channels is estimated and included and the adaptive filters can be updated either with the error signals from all microphones or only from corresponding ones.

Two channel IFXLMS system with all estim

Two channel IFXLMS system with all estimated inverses

A result of the simulations for the above set-up is shown in the animation below. Uncancelled sound signal consisting of five harmonics is shown in black colour. Red coloured trace shows the level of the signal on the one error microphone during the adaptation period of 3 seconds.

Simulation result for two channel ANC se

Simulation result for two channel ANC set-up with all estimated inverses


APPLICATIONS

The potential applications for ANC systems are considerable. They have the potential to reduce the low frequency sound, which is very hard to reduce using conventional methods requiring very heavy and expensive structures. This opens up the whole field of reducing noise from machinery, such as large generators/motors, diesel engines, construction sites, earth moving vehicles, factory machinery and textile mills (most of which now have to operate 24 hours per day to remain competitive which in turn causes noise problems at night).

There is enormous potential socioeconomic benefit in this application. It would help to ameliorate the increasing environmental noise pollution from industry, construction, transportation and domestic noise sources, and help meet new stringent EU noise regulations.


PUBLICATIONS

  1. B. Vuksanovic, D. Nikolic, "Multichannel Active Noise Control in Open Spaces", ICCC 2004, Zakopane, Poland, May 2004.
  2. D. Nikolic, B. Vuksanovic, "Modelling Multichannel ANC Systems in the Freefield Conditions", Inter-Noise 2004, Prague, Chech Republic, August 22-25, 2004.
  3. S. Wright, H. Atmoko, B. Vuksanovic, D. Nikolic: "Portable System for Cancelling Noise in Unconfined Spaces", ACTIVE 04, Williamsburg, Virginia, September 20-22, 2004.
  4. S.E. Wright, H. Atmoko, B. Vuksanovic, "Active control of environmental noise, VIII: increasing the response to primary source changes including unpredictable noise", Journal of Sound and Vibration, vol. 274, pp. 323???349, 2004.