Beamforming with microphone arrays for directional sources

Area of Science:

• Acoustics
• Signal Processing
• Array Signal Processing

Background:

• Conventional beamforming relies on simplified acoustic source models (plane wave or monopole).
• Deviations of real sources from these models degrade beamformer performance, reducing array gain.
• Existing methods struggle with complex or non-ideal acoustic wavefronts.

Purpose of the Study:

• To develop an alternative beamforming approach using a more general source model.
• To enhance array gain and improve the quality of beamforming output for real-world acoustic sources.
• To provide a method for estimating optimal weighting for sub-beamformer outputs.

Main Methods:

• A novel beamforming strategy employing multiple "sub-beamformers" is proposed.
• Each sub-beamformer is tuned to a specific spatial mode of the acoustic source.
• Outputs are combined via a weighted sum, with weights estimated from observed data.

Main Results:

• The proposed method significantly outperforms conventional monopole beamforming.
• Experimental results show an array gain of approximately 11 dB for the new method.
• Conventional beamforming achieved only -4 dB array gain in the same experiment.

Conclusions:

• The generalized source model and sub-beamformer approach effectively addresses limitations of conventional beamforming.
• This method offers superior performance for directive acoustic sources with complex wavefronts.
• The technique demonstrates potential for improved acoustic sensing and source localization in challenging environments.