Classification of Ship Radiated Noise from Recordings Made in the Hudson River
Michael L. Zucker
Stevens Institute of Technology,
Davidson Lab,
Castle Point of the Hudson,
Hoboken, NJ 07030 - (717) 623-0900
Alexander Sedunov
Stevens Institute of Technology,
Davidson Lab,
Castle Point of the Hudson,
Hoboken, NJ 07030
Vladimir Zhdanov
Stevens Institute of Technology,
Davidson Lab,
Castle Point of the Hudson,
Hoboken, NJ 07030
Alexander Sutin - asutin@stevens.edu
Stevens Institute of Technology,
Davidson Lab,
Castle Point of the Hudson,
Hoboken, NJ 07030
Popular version of paper 3aUW9
Presented Wednesday Morning, Oct. 28, 2009
158th ASA Meeting, San Antonio, TX
Given the density of commercial and recreational vessel traffic in many, if not most of the nations estuary and harbor environments, the detection, tracking, and classification of small vessels is required for maritime security. Passive acoustic methods have the ability to quickly and accurately identify small boats and provide invaluable tool is assessing possible threats.
Work is being done at Stevens Institute of Technology to develop methods for extracting useful classification parameters from recordings of radiated ship noise. Through this research assessments of various proposed classification parameters are being conducted on recordings of ship noises made in the Hudson River. The Hudson provides a great environment to collect data of various types of ships due to the heavy shipping traffic encountered. The Center for Maritime Security at Stevens also provides excellent opportunities for both optical (both video and still photography) and acoustic data collection due to the close proximity of the Hudson River to the Stevens campus.
The main sources of radiated ship noise can typically be categorized as one of three classes: 1) mechanical engine noise 2) propeller noise or 3) hydrodynamic noise. In general these signals are mainly composed of some type of amplitude modulated noise, such as the cavitation noise produced by a ships propeller blades. This means that, unlike signals comprised of a manageable set of simple tones, this radiated noise typically possesses random frequency content (much like white noise) that varies in amplitude over time. Thus, some of the most useful signal characteristics are found in the signal's envelope spectrum. Examination of the spectrum of the noise signals envelope (which is often referred to as the signal's DEMON (Detection of Envelope Modulation on Noise) has been shown to be a valuable tool in classifying aspects of a particular ship. The characteristics of the DEMON signal for particular ship are dependent on aspects of that ships performance such as speed, propulsion type and tonnage.
Data collection has been accomplished by recording continuous high resolution acoustic signal from a fixed hydrophone receiver array along with continuous video surveillance recordings made within the same region. This allows researchers to not only extract classification parameters from acoustic signals but to cross reference individual signals with visual observations of each passing ship.
The enclosed video shows the video recording of the Hudson River water traffic together with a spectrogram of the recorded signal and a cross-correlation floating chart. The spectrogram demonstrates the variation of acoustic radiation in the process of the boat movement, and the cross-correlation floating chart can be used for finding the direction to the boat as was described in our other lay-language paper (L.Fillinger, A.Sutin, A.Sedunov. Cross correlation of ship noise for water traffic monitoring).
See the video of Hudson River water traffic HERE.
The following figures show several vessels and their corresponding envelope (DEMON) spectra. Upon first inspection it becomes easy to see that for various types of vessels drastic differences in their envelope spectrums are evident.
The parameters currently studied in each DEMON spectrum include the frequency of the fundamental peak, the modulation index of each peak, the number of peaks in each spectrum, the slope of a linear fit to all peaks in each spectrum and a measure of the signal's total harmonic distortion.
Assessment of these parameters is ongoing, though preliminary results have shown a large variation of these parameters that depends on ship class and speed. Once the usefulness of each parameter is assessed future work will be aimed at the development of a real time classification system based on fast extraction of the most useful classification parameters. Such a system would be an inexpensive yet invaluable tool in assessing threats from nearby ships by identifying their characteristics (vessel size/type, speed and bearing).
This work was supported by the U.S. Department of Homeland Security under Grant Award Number 2008-ST-061-ML0002. The view and conclusions containing in this document are those of the authors and should not be interrelated as necessarily representing the official policies, ether expressed or implied, of the U.S. Department of Homeland Security.