Tracking Changes in the Shallow Water Oceanic Environment Using Geometric Feature Representation and Pattern Learning Techniques
Abstract
The proposed work involves harnessing adaptive signal processing with geometric feature extraction and sophisticated ~big data~ mac"hine learning techniques to track, interpret, and learn potentially intertwined signatures of diverse oceanic events in shallow water acoustics. Oceanic phenomena of interest include but are not limited to multipath acoustic scattering, Doppler effect from fluid motion and moving reflectors in the ocean, high-energy transient events such as surface wave focusing, constructive multipath interference, structured ambient noise, etc. To achieve this end, we will develop novel real-time feature extraction and channel recovery algorithms that springboard off manifold signal processing and sparse representation techniques recently developed by the PI for disentangling potentially overlapped high-energy features in space plasma physics and shallow water acoustic communications. proposed signal processing techniques will provide analytical rigor (e.g. Cramer-Rao-like bounds on channel prediction performance) as well as opportunity to enhance our real-time knowledge of the ocean environment at different scales of time, frequency, space and sparsity using a suite ofsophisticated machine learning techniques. The technical focus of this proposal is for the shallowwater acoustic paradigm, though with some modifications, the computational techniques proposed may apply to deep-water scenarios as well. Our methods will be validated over experimental field data across multiple experiments under diverse oceanic conditions, as well as over exten"sive channel simulations using under diverse environmental constraints.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 20, 2019
- Source ID
- N000141912609
Entities
People
- Ananya Sen Gupta
Organizations
- Office of Naval Research
- United States Navy
- University of Iowa