A fast adaptive algorithm in frequency domain for multichannel feedforward acive noise control system

The reference paths from original sources to reference sensors in multichannel feedforward active noise control (ANC) system are often ignored by most ANC algorithms. A two-staged adaptive algorithm in the frequency domain is proposed to deal with some more complicated cases, especially when the reference sensors cannot be close to noise sources. The two-staged system includes a blind preprocessing system cascaded with a traditional MFxLMS system, and both Subsystems work in the frequency domain. The new algorithm can eliminate the effect of reference paths by blind preprocessing independently and the MFxLMS algorithm in the following stage can obtain faster convergence speed than the traditional one without blind preprocessing. The computational complexity by the proposed algorithm is analyzed and numerical Simulation with impulse responses measured in a real reverberant room is carried out to verify the convergence Performance of the proposed algorithm.