Microphone Subset Selection for MVDR Beamformer Based Noise Reduction

Abstract

In large-scale wireless acoustic sensor networks (WASNs), many of the sensors will only have a marginal contribution to a certain estimation task. Involving all sensors increases the energy budget unnecessarily and decreases the lifetime of the WASN. Using microphone subset selection, also termed as sensor selection, the most informative sensors can be chosen from a set of candidate sensors to achieve a prescribed inference performance. In this paper, we consider microphone subset selection for minimum variance distortionless response (MVDR) beamformer based noise reduction. The best subset of sensors is determined by minimizing the transmission cost while constraining the output noise power (or signal-to-noise ratio). Assuming the statistical information on correlation matrices of the sensor measurements is available, the sensor selection problem for this model-driven scheme is first solved by utilizing convex optimization techniques. In addition, to avoid estimating the statistics related to all the candidate sensors beforehand, we also propose a data-driven approach to select the best subset using a greedy strategy. The performance of the greedy algorithm converges to that of the model-driven method, while it displays advantages in dynamic scenarios as well as on computational complexity. Compared to a sparse MVDR or radius-based beamformer, experiments show that the proposed methods can guarantee the desired performance with significantly less transmission costs.

Rate-Distributed Spatial Filtering Based Noise Reduction in Wireless Acoustic Sensor Networks

“…By doing so, we find the link between rate allocation and sensor selection problems, i.e., rate allocation is a generalization of sensor selection. In [7], the best microphone subset is chosen by minimizing the total transmission costs and constraining the noise reduction performance, where the transmission cost between each node and the FC is only considered as a function of distance. The selected microphone will communicate with the FC using the maximum bit rate.…”

“…The selected microphone will communicate with the FC using the maximum bit rate. The energy model of the approach in the current paper is more general as compared to that in [7]. Based on the rates obtained by the proposed RD-LCMV approach, the best microphone subset of MD-LCMV can be determined by putting a threshold on the rates, e.g., the sensors whose rates are larger than this threshold are chosen.…”

“…We first find that sensor selection is a special case of the rate allocation problem. Then, we propose a bisection algorithm that can be used to obtain the sensor selection results as in [7] based on the rate allocation method.…”

Robust Joint Estimation of Multimicrophone Signal Model Parameters

“…M ICROPHONE arrays (see e.g., [1] for an overview) are used extensively in many applications, such as source separation [2]- [6], multi-microphone noise reduction [1], [7]- [13], dereverberation [14]- [19], sound source localization [20]- [23], and room geometry estimation [24], [25]. All the aforementioned applications are based on a similar multimicrophone signal model, typically depending on the following parameters: i) the early relative acoustic transfer functions (RATFs) of the sources with respect to the microphones; ii) the power spectral densities (PSDs) of the early components of the sources, iii) the PSD of the late reverberation, and, iv) the PSDs of the microphone-self noise.…”

Hybrid Sparse Array Beamforming Design for General Rank Signal Models

“…It is important to note that having fully augmentable arrays not only provide the benefits of simplified sensor switching and improved identifiability of large number of sources, but also they ensure the availability of full array data covariance matrix essential to carry optimized SINR configuration [37], [38]. Therefore, the proposed simplified hybrid sensor switching architecture ensures the knowledge of global data statistics at all times, in contrast to previous efforts in [39]- [41] that sort to optimize data dependent microphone placement viz a viz transmission power. The proposed methodology therein targets a different objective function and primarily relies on local heuristics.…”