Low-Cost One-bit Non-synchronous Measurements for Sound Source Localization

Recent advancements in Micro-Electro-Mechanical System (MEMS) microphone arrays have led to the tendency of adopting the low-cost hardware, which could be more suitable for the applications in the acoustical signal acquisition systems. In this paper, we considered the sound source localization (SSL) problem of one-bit quantized microphone array-based non-synchronous measurements (NSM). With analysis, it is shown that, we could consider the one-bit covariance matrix as a summation of a scaled unquantized covariance matrix and a scaled identity matrix approximately with an unknown scaling parameter. While the influence of the SSL accuracy by using the subspace-based approaches would be acceptable due to the extreme quantization. With the simulation results and root-mean-square error (RMSE) analysis, we demonstrated that the low-cost one-bit NSM method could still provide an accurate localization performance for a couple of speech signal sources. Meanwhile, the RMSE of the one-bit methods influenced by the signal-to-noise ratio (SNR) is comparable with the methods without quantification.