Acoustic Moving Source Localization using Sparse Time Difference of Arrival Measurements

Abstract

Time difference of arrival (TDoA) measurements are common in distributed localization systems, where events are emitted by the tracked source and the detected arrival times at the sensors, deployed at known locations, are used to compute the source location. Since erroneous measurements are common, e.g. due to non-line-of-sight conditions, the sensor fusion must be robust against such outlier measurements. In this paper the consensus-based TDoA fusion, which was proven very successful in various acoustic localization systems, is extended for the case when the source is moving and the event detections at the sensors are not reliable, thus sparse. The derivation of the extended consensus function will be introduced, and the performance is illustrated through simulation examples, showing accuracy around 10cm in realistic scenarios.