A fully distributed method for acoustic localization with sensor networks
Williams, Stephen Michael
The purpose of this work is to present a scalable distributed sensor network capable of localizing acoustic sources in the near-field. Traditional sensor networks employ a central controller, however; centralized data processing in large-scale sensor networks is not always desirable because of the excessive communication and computational complexity it requires. Therefore, a fully distributed localization algorithm is developed in which nodes form groups and collaborate to locate sources. Source locations are determined from time of arrival (TOA) information of the acoustic wave front using Tikhonov regularized inversion (RI). Experimental results validate the use of a distributed localization algorithm, and compare the effectiveness of the RI method to that of traditional matrix solution methods. Additionally, important parameters of the distributed localization algorithm are considered.