Combined beamforming and noise cancellation

This work examines combined beamformer-noise canceller (BF-NC) structures. A new paradigm for adaptive beamforming is presented: The front-end consists of a bank of arbitrary beamformers which are configured independently by a beam-steering module. A switching unit then cascades the beamformers with a multiple-reference adaptive noise canceller which provides sidelobe suppression. The relationship between the front-end beamformers and cascaded adaptive noise cancellers is characterized by computing the beampattem, directivity, white noise gain, and signal-to-interference ratio (SIR) before and after the noise canceller stage. It is shown that the SIR gain offered by the desired signal beamformer varies inversely with the SIR gain contributed by the noise canceller. It is also shown that signal enhancement is offered for a broad range of scenarios, including a speech desired signal and speech interference. An emphasis is placed on evaluating BF-NC structures in real environments. Experimental results reveal SIR improvements of up to 30 dB.