SBIR-STTR Award

Within the increasingly contested undersea operational arena, the Navy needs a tactical and competitive advantage in undersea sensing and detection through improved situational awareness with respect to environmental parameters, such as sound speed profile (SSP) and bottom properties, that affect overall submarine sonar sensor performance. Current approaches for enhancing environmental situational awareness rely heavily on historical databases and remote numerical ocean models to provide predictions of the acoustic environment and sensor performance within that environment. To gain this competitive advantage and enhance the tactical decisions and warfighting posture of submarines, the Navy needs to develop advanced environmental inference capabilities to provide in-situ characterizations of the speed and attenuation of sound in the seabed and water column. For the purposes of this research , HRC proposes to create a Deep Learning Environmental Inference Framework that will utilize platform/sensor/environmental geometry, canonical representations of available passive and active sensors, information derived from these available submarine sonar systems, inference and deep learning algorithms, and standard Navy ocean acoustic models coupled with advanced 3-D acoustic propagation model capabilities to fuse the traditional sources of data with local measurements to improve the currency of the environmental picture and provide measures of uncertainty for derived parameters.

Benefit:
The primary anticipated benefits of advanced sound speed profile and bottom loss inference algorithms and soundfield visualization techniques are to the U.S. Navys submarine community where accurate layer detection in vertical profiles and bottom composition in multipath environments are crucial in avoiding detection as well as providing a competitive edge in detecting potential subsurface threats. However, the surface ASW, P-8 & MH60-R air, and surveillance communities will also benefit from the improved resolution in vertical sound speed profiles and layer detection through enhanced threat detection, improved sensor performance prediction, and sensor placement optimization (especially for multistatic sonobuoy fields in littoral waters). These organizations will also benefit indirectly from another Navy beneficiary of this technology through the Naval Oceanographic Offices (NAVOCEANO) ability to create higher resolution environmental databases. Improved sound speed extraction will allow more accurate historical sound speed databases such as Generalized Digital Environmental Model (GDEM) as well meteorological forecast data as Modular Ocean Data Assimilation System (MODAS), Navy Coastal Ocean Model (NCOM), and Navy Coupled Ocean Data Assimilation (NCODA) Forward formats. Additionally, NAVOCEANO uses the Navy Standard PE model to invert acoustic measurements to estimate low-frequency bottom loss parameters for use in LFBL and HFBL databases. Improvements in bottom loss estimation coupled with a mid-frequency 3D propagation model could be used to greatly enhance the databases. HRCs commercialization strategy involves both government and private uses the within oceanographic and climatology research as well as the oil/gas industry. One possible commercial application of environmental inference algorithms is within climatology. Within conventional climatology, representations of vertical oceanographic profiles are based on mean or median profiles of historic data throughout the world. In areas containing oceanographic fronts, mean profiles may not be representative for the in-situ profiles in the area. Additionally, in areas with highly varying topography, depth- and range-dependent behaviors of all historic temperature and salinity profiles are sparse within current climatology databases and models. Utilizing deep-learning driven environmental inference algorithms coupled with an advanced 3-D propagation model that takes into account highly varying topology may be an approach to generate more realistic climatological estimates of the vertical profiles at a given time and area. Oil and gas industry may also benefit from advanced, wide-area bottom loss inference algorithms in identifying geological bottom composition in searching for oil and gas deposits. Current exploration utilizes sound gun methods that are restricted in their field of resolution (narrow-field systems). As side-scanning sonars become more sophisticated coupled with broad-area bottom loss inference algorithms, advanced 3-D propagation models, and advanced soundfield and bottom loss visualization techniques, exploration ships could gain the ability detect bottom composition over a broader area resulting in larger, better coverage in a shorter period of time, Such an advance could be a huge cost savings in a financial climate in which cutting costs and increasing profits are critical to the survival of a company.

Keywords:
acoustic measurements, acoustic measurements, Acoustics, Sensor Performance, Optimization, data fusion, deep-learning, Environmental Parameters, inference framework

Within the increasingly contested undersea operational arena, the Navy needs a tactical and competitive advantage in undersea sensing and detection through improved situational awareness with respect to environmental parameters, such as sound speed profile (SSP) and bottom properties, that affect overall submarine sonar sensor performance. Current approaches for enhancing environmental situational awareness rely heavily on historical databases and remote numerical ocean models to provide predictions of the acoustic environment and sensor performance within that environment. To gain this competitive advantage and enhance the tactical decisions and warfighting posture of submarines, the Navy needs to develop advanced environmental inference capabilities to provide in-situ characterizations of the speed and attenuation of sound in the seabed and water column. For the purposes of this research , HRC proposes to create a Deep Learning Environmental Inference Framework that will utilize platform/sensor/environmental geometry, canonical representations of available passive and active sensors, information derived from these available submarine sonar systems, inference and deep learning algorithms, and standard Navy ocean acoustic models coupled with advanced 3-D acoustic propagation model capabilities to fuse the traditional sources of data with local measurements to improve the currency of the environmental picture and provide measures of uncertainty for derived parameters.

Benefit:
The primary anticipated benefits of advanced sound speed profile and bottom loss inference algorithms and soundfield visualization techniques are to the U.S. Navys submarine community where accurate layer detection in vertical profiles and bottom composition in multipath environments are crucial in avoiding detection as well as providing a competitive edge in detecting potential subsurface threats. However, the surface ASW, P-8 & MH60-R air, and surveillance communities will also benefit from the improved resolution in vertical sound speed profiles and layer detection through enhanced threat detection, improved sensor performance prediction, and sensor placement optimization (especially for multistatic sonobuoy fields in littoral waters). These organizations will also benefit indirectly from another Navy beneficiary of this technology through the Naval Oceanographic Offices (NAVOCEANO) ability to create higher resolution environmental databases. Improved sound speed extraction will allow more accurate historical sound speed databases such as Generalized Digital Environmental Model (GDEM) as well meteorological forecast data as Modular Ocean Data Assimilation System (MODAS), Navy Coastal Ocean Model (NCOM), and Navy Coupled Ocean Data Assimilation (NCODA) Forward formats. Additionally, NAVOCEANO uses the Navy Standard PE model to invert acoustic measurements to estimate low-frequency bottom loss parameters for use in LFBL and HFBL databases. Improvements in bottom loss estimation coupled with a mid-frequency 3D propagation model could be used to greatly enhance the databases. HRCs commercialization strategy involves both government and private uses the within oceanographic and climatology research as well as the oil/gas industry. One possible commercial application of environmental inference algorithms is within climatology. Within conventional climatology, representations of vertical oceanographic profiles are based on mean or median profiles of historic data throughout the world. In areas containing oceanographic fronts, mean profiles may not be representative for the in-situ profiles in the area. Additionally, in areas with highly varying topography, depth- and range-dependent behaviors of all historic temperature and salinity profiles are sparse within current climatology databases and models. Utilizing deep-learning driven environmental inference algorithms coupled with an advanced 3-D propagation model that takes into account highly varying topology may be an approach to generate more realistic climatological estimates of the vertical profiles at a given time and area. Oil and gas industry may also benefit from advanced, wide-area bottom loss inference algorithms in identifying geological bottom composition in searching for oil and gas deposits. Current exploration utilizes sound gun methods that are restricted in their field of resolution (narrow-field systems). As side-scanning sonars become more sophisticated coupled with broad-area bottom loss inference algorithms, advanced 3-D propagation models, and advanced soundfield and bottom loss visualization techniques, exploration ships could gain the ability detect bottom composition over a broader area resulting in larger, better coverage in a shorter period of time, Such an advance could be a huge cost savings in a financial climate in which cutting costs and increasing profits are critical to the survival of a company.

Keywords:
deep-learning, acoustic measurements, Sensor Performance, Environmental Parameters, Acoustics, Optimization, data fusion, inference framework