Optimization of number of microphones and microphone spacing using time delay based multilateration approach for explosive sound source localization

Abstract

The use of microphone array for sound source localization has been a trending area of research in speech source recognition, battlefield acoustics, etc. For such applications, the localization accuracy is significantly influenced by the number of microphones and their arrangement on the ground surface. Thus, this paper proposes optimized microphone array geometries with the optimized number of microphones for positioning near and distant sound sources. Moreover, this paper also discusses the significance of the reference microphone selection on positioning accuracy. Here, the time-delay-based multilateration approach is used to formulate the governing nonlinear hyperbolic equations for sound source localization. The resulting equations are solved using the least square optimizer, the modified Levenberg Marquardt algorithm. The source results for several customized array structures are validated through simulation and experimental studies. Based on the results, this paper proposes that at least six microphones are required to identify a source position (<1800 m) with 97 % accuracy at lower noise thresholds (SNR: (6 to 20) dB) and 95 % accuracy at higher noise thresholds (SNR: -6 to 0 dB). Moreover, the paper suggests that microphone spacing should be a minimum of 0.34 times the source range for localization of distant sound sources to improve accuracy at higher and lower noise levels.