Comparison of Microphone Array Processing Techniques for Aeroacoustic Measurements

Abstract

This paper presents a systematic comparison of several prominent beamforming algorithms developed for aeroacoustic measurements. The most widely used delay-and-sum (DAS) beamformer is known to suffer from high sidelobe level and low resolution problems. Therefore, more advanced methods, in particular the deconvolution approach for the mapping of acoustic sources (DAMAS), sparsity constrained DAMAS (SC-DAMAS), covariance matrix fitting (CMF) and CLEAN based on spatial source coherence (CLEAN-SC), have been considered to achieve improved resolution and more accurate signal power estimates. The performances of the aforementioned algorithms are evaluated via experiments involving a 63-element logarithmic spiral microphone array in the presence of a single source, two incoherent sources with similar strengths and with different strengths, and two coherent sources. It is observed that DAMAS, SC-DAMAS and CMF provide the most reliable source location estimates, even at relatively low frequencies. Furthermore, the integrated levels obtained with the array processing algorithms are shown to agree with what a single reference microphone placed at the center of the array measures when the array is appropriately calibrated. It is also shown that, as expected, the aforementioned algorithms are unsuccessful in distinguishing coherent acoustic sources unless the frequency is relatively high. DAS and CLEAN-SC are shown to be around 2 to 90 times faster than the other three algorithms.