38 research outputs found
Near real-time detection of low-frequency baleen whale calls from an autonomous surface vehicle: implementation, evaluation, and remaining challenges
No full text© The Author(s), 2021. This article is distributed under the terms of the Creative Commons Attribution License. The definitive version was published in Baumgartner, M. F., Ball, K., Partan, J., Pelletier, L., Bonnell, J., Hotchkin, C., Corkeron, P. J., & Van Parijs, S. M. Near real-time detection of low-frequency baleen whale calls from an autonomous surface vehicle: implementation, evaluation, and remaining challenges. Journal of the Acoustical Society of America, 149(5), (2021): 2950-2962, https://doi.org/10.1121/10.0004817.Mitigation of threats posed to marine mammals by human activities can be greatly improved with a better understanding of animal occurrence in real time. Recent advancements have enabled low-power passive acoustic systems to be integrated into long-endurance autonomous platforms for persistent near real-time monitoring of marine mammals via the sounds they produce. Here, the integration of a passive acoustic instrument capable of real-time detection and classification of low-frequency (LF) tonal sounds with a Liquid Robotics wave glider is reported. The goal of the integration was to enable monitoring of LF calls produced by baleen whales over periods of several months. Mechanical noises produced by the platform were significantly reduced by lubricating moving parts with polytetrafluoroethylene, incorporating rubber and springs to decelerate moving parts and shock mounting hydrophones. Flow noise was reduced with the development of a 21-element hydrophone array. Surface noise produced by breaking waves was not mitigated despite experimentation with baffles. Compared to a well-characterized moored passive acoustic monitoring buoy, the system greatly underestimated the occurrence of sei, fin, and North Atlantic right whales during a 37-d deployment, and therefore is not suitable in its current configuration for use in scientific or management applications for these species at this time.Funding for this project was provided by the Environmental Security Technology Certification Program of the U.S. Department of Defense and the U.S. Navy's Living Marine Resources Program
Experimentation and at-sea testing of underwater network protcols via software reuse from ns2-Miracle network simulations. A case of study using WHOI Micromodems
No full textThe goal of this thesis is to create a C++ module to interface the ns2/NS-Miracle network simulator with the Micromodem developed by the Woods Hole Oceanographic Institution (WHOI Micromodem).
From the network simulator's point of view, the implemented module acts like a common physical layer, but, rather than to be connected to a simulated channel, it opens a serial connection between the machine running ns2 and the Micromodem. Then, the modem transmits acoustically the packet over the real underwater channel.
In collaboration with the Woods Hole Oceanographic Institution (WHOI), a leader in the field of underwater technology, we created a network testbench composed of seven easy deployable nodes. These nodes are intended to be used for a wide range of applications thanks to their multi purpose configuration: the embedded system is composed by a Gumstix board running Emdebian and it is possible to access each single node of the network through a WiFi SSH connection in order to schedule tests, launch programs and collect resultsope
A Survey of Practical Issues in Underwater Networks
No full textunderwater acoustic networks, mobile disruption-tolerant network
Automatic detection of tropical fish calls recorded on moored acoustic recording platforms
No full textThe deep convolutional networks for the classification of multi-class arrhythmia
No full textAn arrhythmia is an irregular heartbeat. Many researchers in the AI field have carried out the automatic classification of arrhythmias, and the issue that has been widely discussed is imbalanced data. A popular technique for overcoming this problem is the synthetic minority oversampling technique (SMOTE) technique. In this paper, the author adds some sampling of data obtained from other datasets into the primary dataset. In this case, the main dataset is the Massachusetts Institute of Technology–Beth Israel Hospital (MIT-BIH) arrhythmia database and an additional dataset from the MIT-BIH supraventricular arrhythmia database. The classification process is carried out with one-dimensional convolutional neural network model (1D-CNN) to perform multiclass and subject-class advancement of medical instrumentation (AAMII) classifications. The results obtained from this study are an accuracy of 99.10% for multiclass and 99.25% for subject-class
