The biosonar of the Mediterranean Bottlenose dolphins: analysis and modelling of echolocation signals

Abstract

In this chapter, we described the analysis carried out on the sonar clicks emitted by Mediterranean Bottlenose dolphins in both audio and ultrasonic bands. We found that most of the sonar clicks emitted when the dolphin is in front of the hydrophone can be modelled by exponential or by Gaussian broadband multicomponent signals. The parameters of these two models have been estimated. The components characterizing each pulse are generally the first or the first two most powerful ones and the fitting with the data seems to be very good in both audio and ultrasonic band. Actually, the meaning of the sonar clicks in the audio band signals is not clear. Maybe, as reported by Zimmer [20], they can be “machinery noise”, that is noise produced by dolphins in emitting the ultrasonic pulses used for the echolocation. In ultrasonic band the most powerful frequency component is located around 24 kHz, almost four octaves under the frequency peak measured for the Oceanic Bottlenose dolphins. This phenomenon can be mainly due to the differences in the Oceanic and Mediterranean environments. However, the efficiency of the dolphin sonar is not only due to the broadband characteristic of the sonar click signals, then to the very high range resolution. Dolphins are able to use as well a multi-perspective, multi-waveforms approach to sense the targets, moving around their preys and changing their trajectories [21], the power and the Pulse Repetition Frequency (PRF) based upon what they have learned from the previous sonar clicks [11,22]. Finally, they use their trained brain to control the whole biosonar process allowing for versatility and continuous learning [4]. It is, of course, not trivial to build a technological sonar similar to that of the marine mammals but the idea of cognitive radars (and so sonars) has been already proposed in 2006 by Haykin in [23] and some progress has been made since then along that path.