A Low-Cost Robust Distributed Linearly Constrained Beamformer for Wireless Acoustic Sensor Networks with Arbitrary Topology

摘要

We propose a new robust distributed linearly constrained beamformer (BF)which utilizes a set of linear equality constraints to reduce the cross powerspectral density matrix to a block-diagonal form. The proposed BF has aconvenient objective function for use in arbitrary distributed networktopologies while having identical performance to a centralized implementation.Moreover, the new optimization problem is robust to steering vector mismatches(SVMs) and to voice activity detection errors. Two variants of the proposed BFare presented and evaluated in the context of multi-microphone speechenhancement in a wireless acoustic sensor network, and are compared with otherstate-of-the-art distributed BFs in terms of communication costs and robustnessto SVMs.