The alignment problem for bat biosonar beampatterns

Abstract

Unlike the beampatterns of technical acoustical systems, the biosonar beampatterns of bats are highly variable in the shapes of their main-and sidelobes over frequency. Some of this variability could represent adaptations to different sensing tasks. In order to understand such possible adaptations, a quantitative method for the analysis of variability (e.g., principal component analysis) is needed. Since the orientation of biosonar beampatterns is highly variable in-vivo, e.g., due to ear/head movements, and not preserved in isolated noseleaf/ear samples, orientation is left out of the initial analysis. Instead, beampatterns should be aligned to characterize their orientation-independent features. For this purpose, a framework to characterize the beampattern alignment problem and perform the alignment has been drawn up. For each frequency, beampatterns are compared using a distance metric (e.g., a p-norm). By investigating the value of this distance metric over the space of all possible beampattern rotations, it is possible to gain insights into the alignment problem, e.g., with regard to the existence of multiple minima in the metric. This space can also be used to test alignment strategies across multiple frequencies, e.g., trough a weighted sum of the respective distances. © 2013 Acoustical Society of America.