TDOA-Based Acoustic Direction Finding

Introduction

Source localization and Direction of Arrival (DOA) estimation using sensor arrays has drawn significant attention from both academia and industry in the array signal processing field for decades. DOA estimation is usually simply called direction finding. If the source is in the far field of a receiving array, the source signal wavefront arriving at the array can be assumed to be plane wave. Then the DOA can be accurately estimated, while the range estimation error and hence position estimation error would be large especially for the high range-to-baseline ratio case. Thus multiple subarrays are needed for accurate position estimation. For instance, in the two-dimensional case, the crossing of bearing lines produced by two subarrays can be used to estimate the source location.

Many high resolution algorithms such as Multiple Signal Classification (MUSIC) and Estimating Signal Parameters via Rotational Invariance Techniques (ESPRIT) have been considered as the popular beamforming methods for DOA estimation. In contrast to the beamforming approach where the data are processed in the frequency domain, the Time Difference of Arrival (TDOA) based DOA estimator performs the data processing in the time domain. In the case where the acoustic signal is transient and does not have a pre-defined waveform pattern, the signal characteristics are unknown such that only the relative time delays can be measured and utilized. The azimuth and elevation can be directly calculated from the estimated time delays so that it does not need a time-consuming two dimensional search over the array manifold as the beamformer does.

The TDOA technique is one of the widely used localization schemes in cellular networks and wireless sensor networks including the use of acoustic arrays. Some indoor localization systems use ultrasonic transceivers to deliver the localization solution with a high accuracy based on the TDOA technique (Yu et al 2009). In a typical scenario where the source is surrounded by a number of sensors (or receivers), a single acoustic target (source) emits a signal, and the sensors detect the signal and measure the arrival time of the source signal. By calculating the signal TDOA at a number of receivers with respect to a reference receiver, the source location can be determined.

Military operations consider the accurate location of weapon system as a vital factor on the battlefield by using relative time of arrival (TOA) of the sound impulse at several accurately located microphones. The relative TOA is commonly referred to as TDOA or simply time delay. The firing of a gun or mortar generates an acoustic impulse, which propagates radially outwards from the source. The TDOA measurements by an array are used to determine the source bearing, and the battlefield target location is cross-fixed by triangulation using the bearing estimates from multiple widely separated arrays (Sallai et al 2011; Brian et al 2002).

Figure 1 illustrates the scenario where an acoustic array is employed to detect and locate the acoustic signal from a single explosion source. The sound source generates a far field signal traveling as an approximate plane wave. In the past the long baseline scheme has been used dominantly to localize a transient explosion source, because the precision of data acquisition systems was not high enough to distinguish the tiny TDOA measurements associated with small distance differences. However, the performance of the present-day equipment has been improved significantly to enable the measurement of the small TDOA values, which allows practical estimation of DOA of far field sources with a number of TDOA measurements.

Figure 1.

Simple schematic of detection of DOA with an acoustic signal by a sensor array