Underwater Acoustic Detection and Localization
Description of work
Our solution will be based on the emission of acoustic sound and the analysis of received reflected templates of the signals from an acoustic array. The data processing will involve spatial signal processing to form adaptive acoustic beams. To face the challenge of false alarms due to the omnipresent acoustic noise transients, we would develop and implement a novel technique for threshold-less acoustic detection of the acoustic reflections that will combine blind clustering with prior scientific knowledge about the motion of the tracked fish. Besides detection, our system will also classify the fish type based on the density and the strength of the received acoustic reflections.
The tasks identified with this WP are:
2.1) Fish Characterization: Lead partner: UH. Determine the expected size, texture, and motion characteristics of the tracked pelagic fish. This task will involve a survey of the specific features of the tracked fish. The conclusions of this survey will be used in Task 2.2 (acoustic detection). The work will be managed by a marine biologist from UH, and will involve IMDEA to learn the capabilities of the acoustic detection and localisation systems. The milestone of this task is the completion of a guide of expected features of the tracked pelagic fish types.
2.2) Detection: Lead partner: UH. This task will involve a design of an algorithm for threshold-less detection of fish based on received acoustic reflections. The work will include forming an adaptive beamforming from the acoustic array hydrophones to mitigate ambient noise, and extracting features from the reflected template of the acoustic emissions like Doppler shift, size of reflector, and time-varying change to distinguish between reflections from the sea boundaries and from a moving object based on Task 2.1. The outcome of this task will provide a statistical analysis of performance based on simulative models of the pelagic fish and experimental measurements. The work will be managed by an acoustic signal-processing expert from UH, and will involve the PI from EvoLogics to guide the target complexity of the developed algorithm to allow real-time implementation on-board the processing tools of the prototype. The milestone of this task is the demonstration at sea of detection of the tracked pelagic fish types.
2.3) Interference cancellation: Lead partner: UH. Design a signal processing technique to mitigate strong surface noise interference caused by the movements of the surface buoy. The task will involve the development of an adaptive filter that will learn the characteristics of correlative features in the ambient noise to identify and remove the interfering noise. The work will be performed by UH, and will involve EvoLogics for technical support during its to implementation of the algorithm as part on-board of EvoLogics acoustic modem hardware in addition to the EvoLogics acoustic modem firmware. The milestone of this task is a statistical analysis showing noise mitigation of more than 20 dB.
2.4) Localisation: Lead partner: IMDEA. As part of this task, an algorithm to estimate the range and bearing of the detected fish or school of fish from a single location will be developed. The algorithm will involve merging of bathymetric information from the three mooring sites, path estimation of the reflected signals, and spatial processing of the signals received in the acoustic array for bearing estimation. The developed algorithm will be tested in numerical simulations, and in four sea experiments before the integration over the prototype system. The outcome of this work will serve to trigger the corresponding optical camera (as part of WP3). The work will be managed by an acoustic localization expert from IMDEA, and will involve an acoustic signal-processing expert from UH for the acoustic spatial signal processing. The milestone of this task is the demonstration at sea of an accurate ranging and bearing estimation of a detected fish.
2.5) Acoustic Classification: Lead partner: UH. In this task, a mechanism will be developed to distinguish between the detection of predator fish and other pelagic fish types, and to estimate the biomass of the detected fish. The work will involve a technique to identify the acoustic path arrivals from the detected pelagic fish; a method to determine the acoustic target strength and the resonance frequency of the reflections identified with the detected pelagic fish, and to match it with the expected ones of the tracked fish species; a hyper-resolution method to quantify the path arrivals from fish individuals; and a fuzzy logic mechanism to fuse all the measured data for classification and biomass evaluation. The outcome of this task will be telemetered to the shore (as part of WP4), and will serve as an initial point for the optical classification process (as part of WP3). The work will be managed by a geophysics expert from UH, and will involve an acoustic signal-processing expert from UH for the identification and quantification of fish-related reflected paths, and an image processing expect from UH to integrate the outcome of the acoustic classification with the optical classification. The milestone of this task is the demonstration at sea of an accurate identification of a tracked pelagic fish.