Modelling simultaneous echo waveform reconstruction and localization in bats

Abstract

Echolocating bats perceive the world through sound signals reflecting from the objects around them. In these signals, information is contained about reflector location and reflector identity. Bats are able to extract and separate the cues for location from those that carry identification information. We propose a model based on Wiener deconvolution that also performs this separation for a virtual system mimicking the echolocation system of the lesser spearnosed bat, Phyllostomus discolor. In particular, the model simultaneously reconstructs the reflected echo signal and localizes the reflector from which the echo originates. The proposed technique is based on a model that performs a similar task based on information from the frog’s lateral line system. We show that direct application of the frog model to the bat sonar system is not feasible. However, we suggest a technique that does apply to the bat biosonar and indicate its performance in the presence of noise.